diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,70824 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 11799,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.649717514124294e-08,
+      "loss": 1.3884,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1299435028248588e-07,
+      "loss": 1.378,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1299435028248588e-07,
+      "loss": 1.4577,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6949152542372883e-07,
+      "loss": 1.4417,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2598870056497177e-07,
+      "loss": 1.3803,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2598870056497177e-07,
+      "loss": 1.4491,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8248587570621473e-07,
+      "loss": 1.4094,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3898305084745766e-07,
+      "loss": 1.3807,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3898305084745766e-07,
+      "loss": 1.3895,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9548022598870065e-07,
+      "loss": 1.4583,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5197740112994353e-07,
+      "loss": 1.3578,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.084745762711865e-07,
+      "loss": 1.3159,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.649717514124295e-07,
+      "loss": 1.3613,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.214689265536724e-07,
+      "loss": 1.3607,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.779661016949153e-07,
+      "loss": 1.3289,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.344632768361583e-07,
+      "loss": 1.2978,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.909604519774013e-07,
+      "loss": 1.3121,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.474576271186441e-07,
+      "loss": 1.2525,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.039548022598871e-07,
+      "loss": 1.2964,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.6045197740113e-07,
+      "loss": 1.322,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.016949152542373e-06,
+      "loss": 1.2147,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.073446327683616e-06,
+      "loss": 1.2537,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.129943502824859e-06,
+      "loss": 1.2565,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.186440677966102e-06,
+      "loss": 1.2331,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2429378531073449e-06,
+      "loss": 1.2578,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2994350282485877e-06,
+      "loss": 1.2789,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2994350282485877e-06,
+      "loss": 1.3013,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3559322033898307e-06,
+      "loss": 1.2531,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4124293785310736e-06,
+      "loss": 1.231,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4689265536723166e-06,
+      "loss": 1.2459,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5254237288135596e-06,
+      "loss": 0.6509,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5819209039548026e-06,
+      "loss": 1.2699,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6384180790960452e-06,
+      "loss": 1.2798,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6949152542372882e-06,
+      "loss": 1.2228,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7514124293785311e-06,
+      "loss": 1.2954,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8079096045197741e-06,
+      "loss": 1.2202,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8644067796610171e-06,
+      "loss": 1.2223,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.92090395480226e-06,
+      "loss": 1.2364,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.977401129943503e-06,
+      "loss": 1.2612,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.033898305084746e-06,
+      "loss": 1.2178,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.090395480225989e-06,
+      "loss": 1.2097,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.146892655367232e-06,
+      "loss": 0.5979,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.203389830508475e-06,
+      "loss": 1.1739,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.259887005649718e-06,
+      "loss": 1.1264,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.316384180790961e-06,
+      "loss": 1.1822,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.372881355932204e-06,
+      "loss": 1.2321,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4293785310734468e-06,
+      "loss": 1.256,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4858757062146898e-06,
+      "loss": 1.1798,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5423728813559323e-06,
+      "loss": 1.1769,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5988700564971753e-06,
+      "loss": 1.1465,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6553672316384183e-06,
+      "loss": 1.2273,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7118644067796613e-06,
+      "loss": 1.1872,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7683615819209043e-06,
+      "loss": 1.2092,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8248587570621473e-06,
+      "loss": 1.1545,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8813559322033903e-06,
+      "loss": 1.1995,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9378531073446333e-06,
+      "loss": 1.1736,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9943502824858762e-06,
+      "loss": 1.2397,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0508474576271192e-06,
+      "loss": 1.1834,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1073446327683622e-06,
+      "loss": 1.2427,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.163841807909605e-06,
+      "loss": 1.2051,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2203389830508473e-06,
+      "loss": 1.2305,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2768361581920903e-06,
+      "loss": 1.195,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.1734,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3898305084745763e-06,
+      "loss": 1.1385,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4463276836158193e-06,
+      "loss": 1.1668,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5028248587570623e-06,
+      "loss": 1.1929,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5593220338983053e-06,
+      "loss": 1.1247,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6158192090395483e-06,
+      "loss": 1.1718,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6723163841807913e-06,
+      "loss": 1.1769,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7288135593220342e-06,
+      "loss": 1.1203,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7853107344632772e-06,
+      "loss": 1.1032,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.84180790960452e-06,
+      "loss": 1.1886,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.898305084745763e-06,
+      "loss": 1.2181,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.954802259887006e-06,
+      "loss": 1.1075,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.011299435028249e-06,
+      "loss": 1.1154,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.067796610169492e-06,
+      "loss": 1.0801,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.124293785310734e-06,
+      "loss": 1.1881,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.180790960451978e-06,
+      "loss": 1.2097,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.23728813559322e-06,
+      "loss": 1.1339,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.293785310734464e-06,
+      "loss": 1.1546,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.350282485875706e-06,
+      "loss": 1.2272,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.40677966101695e-06,
+      "loss": 1.1561,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.463276836158192e-06,
+      "loss": 1.142,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.519774011299436e-06,
+      "loss": 1.1114,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.576271186440678e-06,
+      "loss": 1.1469,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.632768361581922e-06,
+      "loss": 1.1178,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.689265536723164e-06,
+      "loss": 1.1339,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.745762711864408e-06,
+      "loss": 1.1301,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.80225988700565e-06,
+      "loss": 1.2225,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8587570621468936e-06,
+      "loss": 1.1336,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.915254237288136e-06,
+      "loss": 1.1137,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9717514124293796e-06,
+      "loss": 1.1196,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.028248587570622e-06,
+      "loss": 1.1061,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.084745762711865e-06,
+      "loss": 1.1832,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.141242937853108e-06,
+      "loss": 1.1381,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.197740112994351e-06,
+      "loss": 1.179,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.254237288135594e-06,
+      "loss": 1.1257,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.310734463276837e-06,
+      "loss": 1.1126,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.36723163841808e-06,
+      "loss": 1.1322,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.423728813559323e-06,
+      "loss": 1.1438,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.480225988700565e-06,
+      "loss": 1.1749,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.536723163841809e-06,
+      "loss": 1.1639,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.593220338983051e-06,
+      "loss": 1.1068,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6497175141242946e-06,
+      "loss": 0.6407,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.706214689265537e-06,
+      "loss": 1.1342,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7627118644067805e-06,
+      "loss": 1.107,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.819209039548023e-06,
+      "loss": 1.1293,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8757062146892665e-06,
+      "loss": 1.0925,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.932203389830509e-06,
+      "loss": 1.107,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9887005649717525e-06,
+      "loss": 1.1006,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.045197740112995e-06,
+      "loss": 1.1213,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.1016949152542385e-06,
+      "loss": 1.1357,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.158192090395481e-06,
+      "loss": 1.1567,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2146892655367244e-06,
+      "loss": 1.17,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.271186440677966e-06,
+      "loss": 1.1304,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.32768361581921e-06,
+      "loss": 1.1451,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.384180790960452e-06,
+      "loss": 1.0918,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.440677966101695e-06,
+      "loss": 1.1228,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.497175141242938e-06,
+      "loss": 1.1693,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.553672316384181e-06,
+      "loss": 1.1009,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.610169491525424e-06,
+      "loss": 1.1036,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.1552,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.72316384180791e-06,
+      "loss": 1.1047,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.779661016949153e-06,
+      "loss": 1.1198,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.836158192090396e-06,
+      "loss": 1.1165,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.892655367231639e-06,
+      "loss": 1.0931,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.949152542372882e-06,
+      "loss": 1.1432,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0056497175141246e-06,
+      "loss": 1.0662,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.062146892655368e-06,
+      "loss": 1.0915,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1186440677966106e-06,
+      "loss": 1.1308,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.175141242937854e-06,
+      "loss": 1.0976,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2316384180790965e-06,
+      "loss": 1.1738,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.288135593220339e-06,
+      "loss": 1.078,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3446327683615825e-06,
+      "loss": 1.1614,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.401129943502825e-06,
+      "loss": 1.0724,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4576271186440685e-06,
+      "loss": 1.1434,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.514124293785311e-06,
+      "loss": 1.0589,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5706214689265545e-06,
+      "loss": 1.0929,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.627118644067797e-06,
+      "loss": 1.1333,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.68361581920904e-06,
+      "loss": 1.0991,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.740112994350283e-06,
+      "loss": 1.1076,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.796610169491526e-06,
+      "loss": 1.0905,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.853107344632768e-06,
+      "loss": 1.103,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.909604519774012e-06,
+      "loss": 1.1253,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.966101694915255e-06,
+      "loss": 1.1252,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.022598870056498e-06,
+      "loss": 1.1281,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.07909604519774e-06,
+      "loss": 1.0945,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.135593220338983e-06,
+      "loss": 1.0634,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.192090395480227e-06,
+      "loss": 1.1374,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.248587570621469e-06,
+      "loss": 1.107,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.305084745762712e-06,
+      "loss": 1.1047,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.361581920903955e-06,
+      "loss": 1.1244,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.418079096045199e-06,
+      "loss": 1.1011,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.47457627118644e-06,
+      "loss": 1.1024,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.531073446327684e-06,
+      "loss": 1.1067,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.587570621468927e-06,
+      "loss": 1.1023,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.64406779661017e-06,
+      "loss": 1.1037,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.700564971751413e-06,
+      "loss": 1.0937,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.757062146892656e-06,
+      "loss": 1.0828,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8135593220339e-06,
+      "loss": 1.0621,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.870056497175143e-06,
+      "loss": 1.1225,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.926553672316384e-06,
+      "loss": 0.493,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.983050847457628e-06,
+      "loss": 1.1269,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.039548022598871e-06,
+      "loss": 1.1084,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.096045197740113e-06,
+      "loss": 0.4502,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.152542372881356e-06,
+      "loss": 1.1291,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.2090395480226e-06,
+      "loss": 1.1301,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.265536723163843e-06,
+      "loss": 1.0792,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.322033898305085e-06,
+      "loss": 1.1163,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.378531073446328e-06,
+      "loss": 1.0728,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.435028248587572e-06,
+      "loss": 1.1042,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.491525423728815e-06,
+      "loss": 1.1105,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.548022598870057e-06,
+      "loss": 1.1154,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.6045197740113e-06,
+      "loss": 1.0794,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.661016949152544e-06,
+      "loss": 1.1082,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.717514124293787e-06,
+      "loss": 1.0852,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.774011299435029e-06,
+      "loss": 1.1215,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.830508474576272e-06,
+      "loss": 1.0512,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.887005649717516e-06,
+      "loss": 1.0278,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.943502824858759e-06,
+      "loss": 1.0776,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.099,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0056497175141244e-05,
+      "loss": 1.0948,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0112994350282486e-05,
+      "loss": 1.0444,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.016949152542373e-05,
+      "loss": 1.0608,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0225988700564973e-05,
+      "loss": 1.1165,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0282485875706216e-05,
+      "loss": 1.0671,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0338983050847458e-05,
+      "loss": 1.0921,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0395480225988701e-05,
+      "loss": 1.0718,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0451977401129945e-05,
+      "loss": 1.1057,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0508474576271188e-05,
+      "loss": 1.1209,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.056497175141243e-05,
+      "loss": 1.1051,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0621468926553673e-05,
+      "loss": 1.093,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0677966101694917e-05,
+      "loss": 1.0366,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.073446327683616e-05,
+      "loss": 1.1563,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0790960451977402e-05,
+      "loss": 1.136,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0847457627118645e-05,
+      "loss": 1.1101,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0903954802259889e-05,
+      "loss": 1.077,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.096045197740113e-05,
+      "loss": 1.0568,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1016949152542374e-05,
+      "loss": 1.1129,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1073446327683617e-05,
+      "loss": 1.0359,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.112994350282486e-05,
+      "loss": 1.0436,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1186440677966102e-05,
+      "loss": 1.0323,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1242937853107346e-05,
+      "loss": 1.1182,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1299435028248589e-05,
+      "loss": 1.0661,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1355932203389833e-05,
+      "loss": 1.0737,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1412429378531074e-05,
+      "loss": 1.079,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1468926553672318e-05,
+      "loss": 1.1136,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1525423728813561e-05,
+      "loss": 1.0072,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1581920903954804e-05,
+      "loss": 1.1166,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1638418079096046e-05,
+      "loss": 1.1226,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.169491525423729e-05,
+      "loss": 1.1115,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1751412429378533e-05,
+      "loss": 1.1034,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1807909604519776e-05,
+      "loss": 1.1018,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1864406779661018e-05,
+      "loss": 1.0878,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1920903954802262e-05,
+      "loss": 1.1117,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1977401129943505e-05,
+      "loss": 1.0623,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2033898305084745e-05,
+      "loss": 1.1102,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.209039548022599e-05,
+      "loss": 1.0959,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2146892655367234e-05,
+      "loss": 1.1273,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2203389830508477e-05,
+      "loss": 1.1254,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2259887005649717e-05,
+      "loss": 1.0387,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2316384180790962e-05,
+      "loss": 1.0618,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2372881355932205e-05,
+      "loss": 0.5558,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2429378531073449e-05,
+      "loss": 1.0926,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2485875706214689e-05,
+      "loss": 1.0787,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2542372881355932e-05,
+      "loss": 1.0949,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2598870056497177e-05,
+      "loss": 1.1051,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.265536723163842e-05,
+      "loss": 1.1403,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2711864406779661e-05,
+      "loss": 1.0802,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2768361581920904e-05,
+      "loss": 1.0838,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.282485875706215e-05,
+      "loss": 1.0778,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.288135593220339e-05,
+      "loss": 1.0564,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2937853107344633e-05,
+      "loss": 1.095,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2994350282485876e-05,
+      "loss": 1.0425,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.305084745762712e-05,
+      "loss": 1.0842,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3107344632768361e-05,
+      "loss": 1.0947,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3163841807909605e-05,
+      "loss": 0.5516,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3220338983050848e-05,
+      "loss": 1.0815,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3276836158192092e-05,
+      "loss": 1.1223,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.0786,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3389830508474577e-05,
+      "loss": 1.1042,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.344632768361582e-05,
+      "loss": 1.0638,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3502824858757064e-05,
+      "loss": 1.0566,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3559322033898305e-05,
+      "loss": 1.1204,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3615819209039549e-05,
+      "loss": 1.0594,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3672316384180792e-05,
+      "loss": 1.1133,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3728813559322034e-05,
+      "loss": 1.0688,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3785310734463277e-05,
+      "loss": 1.0626,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.384180790960452e-05,
+      "loss": 1.0445,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3898305084745764e-05,
+      "loss": 1.0019,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3954802259887006e-05,
+      "loss": 1.0569,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4011299435028249e-05,
+      "loss": 0.9842,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4067796610169493e-05,
+      "loss": 1.1302,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4124293785310736e-05,
+      "loss": 1.0732,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4180790960451978e-05,
+      "loss": 1.1309,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4237288135593221e-05,
+      "loss": 1.0745,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4293785310734465e-05,
+      "loss": 1.0965,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4350282485875708e-05,
+      "loss": 1.0768,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.440677966101695e-05,
+      "loss": 1.0765,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4463276836158193e-05,
+      "loss": 1.0504,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4519774011299436e-05,
+      "loss": 1.1123,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4576271186440678e-05,
+      "loss": 1.059,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4632768361581922e-05,
+      "loss": 1.0884,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4689265536723165e-05,
+      "loss": 1.1305,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4745762711864408e-05,
+      "loss": 1.1714,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.480225988700565e-05,
+      "loss": 1.0549,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4858757062146894e-05,
+      "loss": 1.0536,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4915254237288137e-05,
+      "loss": 1.0342,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.497175141242938e-05,
+      "loss": 1.0724,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5028248587570622e-05,
+      "loss": 1.0713,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5084745762711865e-05,
+      "loss": 1.0566,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5141242937853109e-05,
+      "loss": 1.0664,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5197740112994352e-05,
+      "loss": 1.0919,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5254237288135594e-05,
+      "loss": 0.5584,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5310734463276836e-05,
+      "loss": 1.0846,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.536723163841808e-05,
+      "loss": 1.0591,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5423728813559326e-05,
+      "loss": 1.0922,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5480225988700566e-05,
+      "loss": 1.0548,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.553672316384181e-05,
+      "loss": 1.067,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5593220338983053e-05,
+      "loss": 1.0915,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5649717514124293e-05,
+      "loss": 1.0437,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5706214689265536e-05,
+      "loss": 1.1258,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.576271186440678e-05,
+      "loss": 1.0674,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5819209039548023e-05,
+      "loss": 1.0187,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5875706214689266e-05,
+      "loss": 1.0434,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.593220338983051e-05,
+      "loss": 1.0469,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5988700564971753e-05,
+      "loss": 1.1379,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6045197740112997e-05,
+      "loss": 1.1191,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6101694915254237e-05,
+      "loss": 1.0815,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.615819209039548e-05,
+      "loss": 1.0651,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6214689265536724e-05,
+      "loss": 1.0909,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6271186440677967e-05,
+      "loss": 1.0566,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.632768361581921e-05,
+      "loss": 1.0849,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6384180790960454e-05,
+      "loss": 1.0765,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6440677966101697e-05,
+      "loss": 1.0622,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6497175141242937e-05,
+      "loss": 1.0849,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.655367231638418e-05,
+      "loss": 1.0786,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6610169491525424e-05,
+      "loss": 1.0492,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.0848,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.672316384180791e-05,
+      "loss": 1.0593,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6779661016949154e-05,
+      "loss": 1.0638,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6836158192090398e-05,
+      "loss": 1.0588,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.689265536723164e-05,
+      "loss": 0.6187,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.694915254237288e-05,
+      "loss": 1.0972,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7005649717514125e-05,
+      "loss": 1.0878,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7062146892655368e-05,
+      "loss": 1.1348,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.711864406779661e-05,
+      "loss": 1.0876,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7175141242937855e-05,
+      "loss": 1.0901,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7231638418079098e-05,
+      "loss": 1.0854,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.728813559322034e-05,
+      "loss": 1.098,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.734463276836158e-05,
+      "loss": 1.0851,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7401129943502825e-05,
+      "loss": 1.0673,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.745762711864407e-05,
+      "loss": 1.0693,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7514124293785312e-05,
+      "loss": 1.0784,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7570621468926555e-05,
+      "loss": 1.0681,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.76271186440678e-05,
+      "loss": 1.0522,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7683615819209042e-05,
+      "loss": 1.0305,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7740112994350286e-05,
+      "loss": 1.0547,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7796610169491526e-05,
+      "loss": 1.1239,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.785310734463277e-05,
+      "loss": 1.0379,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7909604519774012e-05,
+      "loss": 1.0792,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7966101694915256e-05,
+      "loss": 1.0776,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.80225988700565e-05,
+      "loss": 1.0039,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8079096045197743e-05,
+      "loss": 1.0744,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8135593220338986e-05,
+      "loss": 1.148,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8192090395480226e-05,
+      "loss": 1.0888,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.824858757062147e-05,
+      "loss": 1.064,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8305084745762713e-05,
+      "loss": 1.0411,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8361581920903956e-05,
+      "loss": 1.0184,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.84180790960452e-05,
+      "loss": 1.0102,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8474576271186443e-05,
+      "loss": 1.0799,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8531073446327686e-05,
+      "loss": 1.0687,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.858757062146893e-05,
+      "loss": 1.0956,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.864406779661017e-05,
+      "loss": 1.0264,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8700564971751413e-05,
+      "loss": 1.045,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8757062146892657e-05,
+      "loss": 1.0962,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.88135593220339e-05,
+      "loss": 1.0917,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8870056497175144e-05,
+      "loss": 1.0987,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8926553672316387e-05,
+      "loss": 0.6663,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.898305084745763e-05,
+      "loss": 1.1022,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9039548022598874e-05,
+      "loss": 1.1091,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9096045197740114e-05,
+      "loss": 1.0293,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9152542372881357e-05,
+      "loss": 1.0346,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.92090395480226e-05,
+      "loss": 1.0795,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9265536723163844e-05,
+      "loss": 1.0589,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9322033898305087e-05,
+      "loss": 0.965,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.937853107344633e-05,
+      "loss": 1.0633,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9435028248587574e-05,
+      "loss": 1.089,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9491525423728814e-05,
+      "loss": 1.0569,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9548022598870058e-05,
+      "loss": 1.0143,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.96045197740113e-05,
+      "loss": 1.0031,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9661016949152545e-05,
+      "loss": 1.1092,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9717514124293785e-05,
+      "loss": 1.045,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.977401129943503e-05,
+      "loss": 1.0513,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9830508474576275e-05,
+      "loss": 1.079,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9887005649717518e-05,
+      "loss": 1.0976,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9943502824858758e-05,
+      "loss": 1.071,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.0709,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999623263108e-05,
+      "loss": 1.0333,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998493052454e-05,
+      "loss": 1.0459,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999996609368124e-05,
+      "loss": 1.0746,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999993972210264e-05,
+      "loss": 1.0524,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999058157907e-05,
+      "loss": 1.0903,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999986437474797e-05,
+      "loss": 1.0662,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999981539897757e-05,
+      "loss": 1.1022,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997588884832e-05,
+      "loss": 1.1023,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999996948432691e-05,
+      "loss": 1.088,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999962326334016e-05,
+      "loss": 1.0889,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995441487017e-05,
+      "loss": 1.0937,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999994574993597e-05,
+      "loss": 1.0477,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999936331532067e-05,
+      "loss": 1.006,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999926159659177e-05,
+      "loss": 1.0754,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999915234318064e-05,
+      "loss": 1.0625,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999903555509546e-05,
+      "loss": 1.0489,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999989112323451e-05,
+      "loss": 1.0609,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999877937493886e-05,
+      "loss": 1.0695,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999863998288675e-05,
+      "loss": 1.0851,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999849305619918e-05,
+      "loss": 1.047,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999983385948873e-05,
+      "loss": 1.0983,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999817659896275e-05,
+      "loss": 1.0853,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999800706843764e-05,
+      "loss": 1.1077,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999783000332486e-05,
+      "loss": 1.0763,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999764540363774e-05,
+      "loss": 1.0536,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999745326939008e-05,
+      "loss": 0.5751,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999725360059648e-05,
+      "loss": 1.0647,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999970463972719e-05,
+      "loss": 1.0909,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999683165943198e-05,
+      "loss": 1.0672,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999966093870929e-05,
+      "loss": 1.0798,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999637958027144e-05,
+      "loss": 1.0661,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999614223898488e-05,
+      "loss": 1.0622,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999958973632511e-05,
+      "loss": 1.0428,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999956449530886e-05,
+      "loss": 1.0656,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999538500851633e-05,
+      "loss": 1.0235,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999951175295539e-05,
+      "loss": 1.0213,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999948425162215e-05,
+      "loss": 1.065,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999945599685398e-05,
+      "loss": 1.0584,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999426988653012e-05,
+      "loss": 1.03,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999939722702143e-05,
+      "loss": 1.0136,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999366711961476e-05,
+      "loss": 1.0637,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999335443475452e-05,
+      "loss": 0.9944,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999303421565714e-05,
+      "loss": 1.1038,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999927064623467e-05,
+      "loss": 1.0405,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99992371174848e-05,
+      "loss": 1.0818,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999202835318612e-05,
+      "loss": 1.0967,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999167799738706e-05,
+      "loss": 1.0586,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999913201074772e-05,
+      "loss": 1.081,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999909546834834e-05,
+      "loss": 1.0303,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999905817254333e-05,
+      "loss": 0.5156,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999020123335496e-05,
+      "loss": 1.0676,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99989813207277e-05,
+      "loss": 1.1222,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998941764722874e-05,
+      "loss": 1.0905,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998901455324e-05,
+      "loss": 1.061,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998860392534105e-05,
+      "loss": 1.0458,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998818576356286e-05,
+      "loss": 1.0309,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998776006793702e-05,
+      "loss": 1.026,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998732683849553e-05,
+      "loss": 1.0262,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998688607527103e-05,
+      "loss": 1.1165,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998643777829674e-05,
+      "loss": 1.0626,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998598194760647e-05,
+      "loss": 1.0142,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998551858323454e-05,
+      "loss": 1.0914,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998504768521588e-05,
+      "loss": 1.0552,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999845692535859e-05,
+      "loss": 1.0781,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999840832883808e-05,
+      "loss": 1.0772,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998358978963702e-05,
+      "loss": 1.0571,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998308875739185e-05,
+      "loss": 1.0619,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998258019168304e-05,
+      "loss": 1.0174,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998206409254886e-05,
+      "loss": 1.0551,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998154046002822e-05,
+      "loss": 1.0949,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998100929416057e-05,
+      "loss": 1.062,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99980470594986e-05,
+      "loss": 1.0468,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997992436254496e-05,
+      "loss": 1.0408,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997937059687872e-05,
+      "loss": 1.0688,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997880929802895e-05,
+      "loss": 1.0601,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99978240466038e-05,
+      "loss": 1.1193,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997766410094868e-05,
+      "loss": 1.072,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999770802028044e-05,
+      "loss": 1.0921,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997648877164922e-05,
+      "loss": 1.0621,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999758898075277e-05,
+      "loss": 1.033,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999752833104849e-05,
+      "loss": 1.0302,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997466928056653e-05,
+      "loss": 1.111,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997404771781893e-05,
+      "loss": 1.0532,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997341862228886e-05,
+      "loss": 0.9923,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997278199402373e-05,
+      "loss": 1.0723,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997213783307153e-05,
+      "loss": 0.9835,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999714861394808e-05,
+      "loss": 1.0727,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997082691330066e-05,
+      "loss": 1.1133,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997016015458072e-05,
+      "loss": 1.0675,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996948586337127e-05,
+      "loss": 0.5354,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999688040397231e-05,
+      "loss": 1.038,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996811468368757e-05,
+      "loss": 1.0724,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996741779531665e-05,
+      "loss": 1.1259,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996671337466283e-05,
+      "loss": 1.0334,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996600142177917e-05,
+      "loss": 1.1036,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996528193671936e-05,
+      "loss": 1.0457,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996455491953755e-05,
+      "loss": 1.0244,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999638203702886e-05,
+      "loss": 1.0518,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999630782890278e-05,
+      "loss": 1.0434,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996232867581104e-05,
+      "loss": 1.0507,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996157153069485e-05,
+      "loss": 1.0705,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996080685373628e-05,
+      "loss": 1.1014,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999600346449929e-05,
+      "loss": 1.0923,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999592549045229e-05,
+      "loss": 1.0507,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995846763238514e-05,
+      "loss": 1.0351,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999576728286388e-05,
+      "loss": 1.0741,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995687049334384e-05,
+      "loss": 1.0912,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999560606265607e-05,
+      "loss": 1.0557,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995524322835035e-05,
+      "loss": 0.4436,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995441829877444e-05,
+      "loss": 1.0802,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995358583789514e-05,
+      "loss": 1.0843,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995274584577512e-05,
+      "loss": 1.0583,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999518983224777e-05,
+      "loss": 1.0456,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995104326806675e-05,
+      "loss": 1.0826,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995018068260667e-05,
+      "loss": 0.9886,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994931056616247e-05,
+      "loss": 1.0827,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999484329187997e-05,
+      "loss": 1.0309,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994754774058448e-05,
+      "loss": 1.0912,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999466550315835e-05,
+      "loss": 0.977,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999457547918641e-05,
+      "loss": 0.9964,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994484702149404e-05,
+      "loss": 1.0616,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999439317205417e-05,
+      "loss": 1.0906,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994300888907613e-05,
+      "loss": 1.0479,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994207852716675e-05,
+      "loss": 1.0847,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994114063488376e-05,
+      "loss": 1.0203,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999401952122978e-05,
+      "loss": 0.5185,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993924225948005e-05,
+      "loss": 1.038,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993828177650237e-05,
+      "loss": 1.0459,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999373137634371e-05,
+      "loss": 1.0757,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993633822035722e-05,
+      "loss": 1.0546,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999353551473362e-05,
+      "loss": 1.0503,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993436454444814e-05,
+      "loss": 1.0369,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999333664117676e-05,
+      "loss": 1.0724,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999323607493699e-05,
+      "loss": 0.9663,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993134755733075e-05,
+      "loss": 1.0565,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999303268357265e-05,
+      "loss": 0.6354,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992929858463405e-05,
+      "loss": 1.0902,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992826280413087e-05,
+      "loss": 1.0731,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992721949429503e-05,
+      "loss": 1.0365,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992616865520515e-05,
+      "loss": 0.9726,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992511028694036e-05,
+      "loss": 1.0588,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992404438958046e-05,
+      "loss": 1.0466,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999229709632057e-05,
+      "loss": 1.05,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99921890007897e-05,
+      "loss": 1.0319,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992080152373587e-05,
+      "loss": 1.074,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991970551080417e-05,
+      "loss": 1.0596,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999186019691846e-05,
+      "loss": 1.0303,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991749089896028e-05,
+      "loss": 1.0774,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999163723002149e-05,
+      "loss": 1.1106,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991524617303282e-05,
+      "loss": 1.0546,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999141125174988e-05,
+      "loss": 1.0215,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999129713336983e-05,
+      "loss": 0.986,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991182262171734e-05,
+      "loss": 0.5206,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991066638164237e-05,
+      "loss": 1.0086,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999095026135606e-05,
+      "loss": 1.0393,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999083313175597e-05,
+      "loss": 1.04,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999071524937279e-05,
+      "loss": 1.024,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990596614215405e-05,
+      "loss": 1.076,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999047722629275e-05,
+      "loss": 1.0663,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990357085613828e-05,
+      "loss": 1.0686,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990236192187682e-05,
+      "loss": 1.0202,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990114546023423e-05,
+      "loss": 1.0583,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998999214713022e-05,
+      "loss": 1.0406,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99898689955173e-05,
+      "loss": 1.0374,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998974509119393e-05,
+      "loss": 1.0352,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998962043416946e-05,
+      "loss": 0.9724,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998949502445327e-05,
+      "loss": 1.0618,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989368862054814e-05,
+      "loss": 1.0165,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989241946983598e-05,
+      "loss": 1.0205,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989114279249185e-05,
+      "loss": 1.0276,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988985858861193e-05,
+      "loss": 1.0488,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988856685829303e-05,
+      "loss": 1.0586,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988726760163246e-05,
+      "loss": 1.0137,
+      "step": 531
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988596081872805e-05,
+      "loss": 1.0239,
+      "step": 532
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988464650967834e-05,
+      "loss": 1.0274,
+      "step": 533
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988332467458233e-05,
+      "loss": 1.0067,
+      "step": 534
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988199531353963e-05,
+      "loss": 1.0892,
+      "step": 535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988065842665036e-05,
+      "loss": 1.0683,
+      "step": 536
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998793140140153e-05,
+      "loss": 1.0445,
+      "step": 537
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987796207573573e-05,
+      "loss": 1.1172,
+      "step": 538
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987660261191353e-05,
+      "loss": 1.0047,
+      "step": 539
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998752356226511e-05,
+      "loss": 1.0378,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987386110805146e-05,
+      "loss": 1.0527,
+      "step": 541
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987247906821817e-05,
+      "loss": 1.0506,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998710895032554e-05,
+      "loss": 1.0758,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998696924132678e-05,
+      "loss": 1.0232,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986828779836065e-05,
+      "loss": 0.9947,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998668756586398e-05,
+      "loss": 1.0602,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998654559942116e-05,
+      "loss": 1.0727,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998640288051831e-05,
+      "loss": 1.0543,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986259409166177e-05,
+      "loss": 1.0338,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986115185375576e-05,
+      "loss": 1.0643,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985970209157367e-05,
+      "loss": 1.0515,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985824480522482e-05,
+      "loss": 1.0657,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985677999481898e-05,
+      "loss": 1.0341,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985530766046646e-05,
+      "loss": 0.5395,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985382780227826e-05,
+      "loss": 1.0485,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985234042036588e-05,
+      "loss": 1.0223,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998508455148414e-05,
+      "loss": 1.0085,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998493430858174e-05,
+      "loss": 1.0614,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984783313340715e-05,
+      "loss": 1.0685,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998463156577244e-05,
+      "loss": 1.1058,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998447906588835e-05,
+      "loss": 1.0883,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998432581369993e-05,
+      "loss": 1.0947,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998417180921873e-05,
+      "loss": 1.0389,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984017052456358e-05,
+      "loss": 0.9766,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983861543424467e-05,
+      "loss": 1.0669,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998370528213478e-05,
+      "loss": 1.0898,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998354826859907e-05,
+      "loss": 1.0288,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983390502829168e-05,
+      "loss": 0.5078,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983231984836958e-05,
+      "loss": 1.0311,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983072714634385e-05,
+      "loss": 1.0349,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982912692233455e-05,
+      "loss": 1.0993,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998275191764622e-05,
+      "loss": 1.0268,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982590390884787e-05,
+      "loss": 1.0891,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998242811196134e-05,
+      "loss": 1.0376,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99822650808881e-05,
+      "loss": 1.0216,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998210129767735e-05,
+      "loss": 1.0348,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981936762341438e-05,
+      "loss": 1.0466,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981771474892752e-05,
+      "loss": 1.0509,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998160543534375e-05,
+      "loss": 1.0185,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998143864370694e-05,
+      "loss": 1.0545,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998127109999489e-05,
+      "loss": 1.0176,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981102804220232e-05,
+      "loss": 1.0216,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980933756395635e-05,
+      "loss": 1.0284,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980763956533842e-05,
+      "loss": 1.0428,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980593404647646e-05,
+      "loss": 0.9525,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99804221007499e-05,
+      "loss": 1.0202,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980250044853507e-05,
+      "loss": 1.0226,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980077236971435e-05,
+      "loss": 1.0784,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979903677116705e-05,
+      "loss": 1.0413,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979729365302387e-05,
+      "loss": 0.4222,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979554301541624e-05,
+      "loss": 1.0466,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99793784858476e-05,
+      "loss": 1.0385,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979201918233567e-05,
+      "loss": 1.0232,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979024598712827e-05,
+      "loss": 1.0437,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997884652729874e-05,
+      "loss": 1.0334,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997866770400472e-05,
+      "loss": 1.0301,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997848812884425e-05,
+      "loss": 1.1003,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978307801830855e-05,
+      "loss": 0.9672,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978126722978117e-05,
+      "loss": 0.5348,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997794489229969e-05,
+      "loss": 1.0484,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977762309809266e-05,
+      "loss": 1.0705,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977578975520605e-05,
+      "loss": 1.0408,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977394889447526e-05,
+      "loss": 1.011,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997721005160389e-05,
+      "loss": 1.0829,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977024462003632e-05,
+      "loss": 1.0095,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997683812066073e-05,
+      "loss": 1.0408,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997665102758923e-05,
+      "loss": 1.0547,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997646318280322e-05,
+      "loss": 0.9998,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976274586316866e-05,
+      "loss": 1.0595,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997608523814437e-05,
+      "loss": 1.0377,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975895138299998e-05,
+      "loss": 1.0693,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997570428679808e-05,
+      "loss": 1.0768,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975512683652985e-05,
+      "loss": 1.0581,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975320328879163e-05,
+      "loss": 1.0339,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99751272224911e-05,
+      "loss": 1.0206,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974933364503347e-05,
+      "loss": 1.0519,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974738754930512e-05,
+      "loss": 1.0032,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997454339378726e-05,
+      "loss": 1.0485,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974347281088305e-05,
+      "loss": 1.085,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997415041684843e-05,
+      "loss": 1.0478,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997395280108246e-05,
+      "loss": 0.9971,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973754433805294e-05,
+      "loss": 1.0781,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973555315031873e-05,
+      "loss": 1.0528,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973355444777205e-05,
+      "loss": 1.0658,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973154823056343e-05,
+      "loss": 1.0133,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972953449884407e-05,
+      "loss": 0.9848,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972751325276568e-05,
+      "loss": 1.0058,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997254844924806e-05,
+      "loss": 1.0194,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972344821814163e-05,
+      "loss": 0.5743,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997214044299023e-05,
+      "loss": 1.06,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971935312791646e-05,
+      "loss": 1.0481,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997172943123388e-05,
+      "loss": 1.0362,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971522798332436e-05,
+      "loss": 1.045,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971315414102886e-05,
+      "loss": 1.0265,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971107278560857e-05,
+      "loss": 1.0538,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997089839172203e-05,
+      "loss": 1.06,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997068875360215e-05,
+      "loss": 1.0566,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970478364217e-05,
+      "loss": 1.0948,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997026722358244e-05,
+      "loss": 1.0676,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970055331714383e-05,
+      "loss": 1.0243,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969842688628792e-05,
+      "loss": 1.0973,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969629294341685e-05,
+      "loss": 1.0175,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996941514886914e-05,
+      "loss": 1.0229,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969200252227298e-05,
+      "loss": 0.9877,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968984604432348e-05,
+      "loss": 0.9806,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968768205500537e-05,
+      "loss": 1.0412,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968551055448172e-05,
+      "loss": 1.0534,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968333154291615e-05,
+      "loss": 1.0495,
+      "step": 648
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968114502047285e-05,
+      "loss": 1.0306,
+      "step": 649
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967895098731657e-05,
+      "loss": 1.0525,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996767494436126e-05,
+      "loss": 1.0124,
+      "step": 651
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996745403895268e-05,
+      "loss": 1.0215,
+      "step": 652
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967232382522564e-05,
+      "loss": 1.014,
+      "step": 653
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996700997508762e-05,
+      "loss": 0.5513,
+      "step": 654
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966786816664595e-05,
+      "loss": 1.0218,
+      "step": 655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996656290727031e-05,
+      "loss": 1.0808,
+      "step": 656
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966338246921633e-05,
+      "loss": 1.0113,
+      "step": 657
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966112835635493e-05,
+      "loss": 1.046,
+      "step": 658
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996588667342888e-05,
+      "loss": 1.0271,
+      "step": 659
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965659760318822e-05,
+      "loss": 1.0933,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965432096322423e-05,
+      "loss": 1.0607,
+      "step": 661
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996520368145684e-05,
+      "loss": 1.0588,
+      "step": 662
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964974515739275e-05,
+      "loss": 1.0145,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964744599187006e-05,
+      "loss": 1.077,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964513931817348e-05,
+      "loss": 1.043,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964282513647686e-05,
+      "loss": 1.03,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964050344695454e-05,
+      "loss": 1.0109,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963817424978144e-05,
+      "loss": 1.0118,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996358375451331e-05,
+      "loss": 1.0627,
+      "step": 669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963349333318557e-05,
+      "loss": 0.9902,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963114161411544e-05,
+      "loss": 1.0167,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996287823881e-05,
+      "loss": 1.0194,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962641565531694e-05,
+      "loss": 1.0215,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962404141594457e-05,
+      "loss": 1.0187,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962165967016185e-05,
+      "loss": 1.0409,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961927041814818e-05,
+      "loss": 0.9707,
+      "step": 676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996168736600836e-05,
+      "loss": 1.0558,
+      "step": 677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996144693961487e-05,
+      "loss": 1.0639,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996120576265247e-05,
+      "loss": 1.0361,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960963835139317e-05,
+      "loss": 1.0337,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960721157093652e-05,
+      "loss": 1.0231,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960477728533756e-05,
+      "loss": 1.0371,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960233549477973e-05,
+      "loss": 0.9941,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99599886199447e-05,
+      "loss": 0.9994,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959742939952393e-05,
+      "loss": 1.0727,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995949650951956e-05,
+      "loss": 0.5474,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995924932866477e-05,
+      "loss": 1.0363,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995900139740665e-05,
+      "loss": 1.0634,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958752715763875e-05,
+      "loss": 1.0128,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958503283755194e-05,
+      "loss": 1.0912,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958253101399388e-05,
+      "loss": 1.0186,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958002168715312e-05,
+      "loss": 1.04,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995775048572188e-05,
+      "loss": 1.04,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957498052438046e-05,
+      "loss": 1.041,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957244868882836e-05,
+      "loss": 0.9803,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956990935075322e-05,
+      "loss": 1.0832,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956736251034643e-05,
+      "loss": 1.0431,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995648081677999e-05,
+      "loss": 1.0269,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956224632330596e-05,
+      "loss": 1.0365,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955967697705782e-05,
+      "loss": 1.0062,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955710012924893e-05,
+      "loss": 0.9744,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955451578007353e-05,
+      "loss": 1.0253,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955192392972628e-05,
+      "loss": 1.0164,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954932457840256e-05,
+      "loss": 1.028,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995467177262981e-05,
+      "loss": 0.5819,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954410337360945e-05,
+      "loss": 0.9718,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995414815205335e-05,
+      "loss": 1.0621,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953885216726788e-05,
+      "loss": 1.0281,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995362153140106e-05,
+      "loss": 1.0435,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953357096096047e-05,
+      "loss": 1.0669,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995309191083166e-05,
+      "loss": 1.0328,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995282597562789e-05,
+      "loss": 1.0208,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952559290504773e-05,
+      "loss": 1.045,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952291855482396e-05,
+      "loss": 0.9955,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952023670580915e-05,
+      "loss": 1.0401,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951754735820542e-05,
+      "loss": 1.0396,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951485051221532e-05,
+      "loss": 1.0714,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951214616804203e-05,
+      "loss": 1.0039,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950943432588942e-05,
+      "loss": 0.9958,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950671498596176e-05,
+      "loss": 1.0121,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950398814846396e-05,
+      "loss": 1.0694,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950125381360144e-05,
+      "loss": 0.9958,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994985119815803e-05,
+      "loss": 1.0585,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949576265260708e-05,
+      "loss": 1.0642,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994930058268889e-05,
+      "loss": 1.013,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994902415046336e-05,
+      "loss": 1.0349,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994874696860493e-05,
+      "loss": 1.0206,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948469037134496e-05,
+      "loss": 1.0539,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948190356073e-05,
+      "loss": 0.9972,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947910925441435e-05,
+      "loss": 1.0333,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947630745260855e-05,
+      "loss": 1.0026,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947349815552373e-05,
+      "loss": 0.9862,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994706813633716e-05,
+      "loss": 1.0642,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946785707636436e-05,
+      "loss": 1.0332,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946502529471478e-05,
+      "loss": 1.066,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946218601863626e-05,
+      "loss": 0.9977,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945933924834277e-05,
+      "loss": 0.9974,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945648498404875e-05,
+      "loss": 0.9905,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945362322596926e-05,
+      "loss": 1.0204,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945075397431997e-05,
+      "loss": 1.0223,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944787722931704e-05,
+      "loss": 1.1003,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944499299117724e-05,
+      "loss": 1.0836,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994421012601179e-05,
+      "loss": 1.0527,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943920203635684e-05,
+      "loss": 1.0062,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994362953201126e-05,
+      "loss": 0.9874,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943338111160413e-05,
+      "loss": 0.9884,
+      "step": 746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943045941105102e-05,
+      "loss": 1.0378,
+      "step": 747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994275302186734e-05,
+      "loss": 1.0627,
+      "step": 748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942459353469202e-05,
+      "loss": 1.0396,
+      "step": 749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942164935932814e-05,
+      "loss": 1.0159,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994186976928036e-05,
+      "loss": 1.0206,
+      "step": 751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941573853534076e-05,
+      "loss": 1.034,
+      "step": 752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994127718871626e-05,
+      "loss": 0.9904,
+      "step": 753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940979774849264e-05,
+      "loss": 0.9958,
+      "step": 754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940681611955502e-05,
+      "loss": 0.9746,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940382700057437e-05,
+      "loss": 1.0263,
+      "step": 756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940083039177594e-05,
+      "loss": 0.9988,
+      "step": 757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939782629338543e-05,
+      "loss": 1.0254,
+      "step": 758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993948147056293e-05,
+      "loss": 1.0462,
+      "step": 759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939179562873437e-05,
+      "loss": 1.04,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993887690629282e-05,
+      "loss": 1.0441,
+      "step": 761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993857350084388e-05,
+      "loss": 1.0228,
+      "step": 762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938269346549473e-05,
+      "loss": 1.042,
+      "step": 763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937964443432528e-05,
+      "loss": 1.0102,
+      "step": 764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937658791516005e-05,
+      "loss": 0.9477,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937352390822945e-05,
+      "loss": 1.002,
+      "step": 766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9937045241376426e-05,
+      "loss": 1.0313,
+      "step": 767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936737343199596e-05,
+      "loss": 0.9864,
+      "step": 768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936428696315656e-05,
+      "loss": 1.0512,
+      "step": 769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993611930074786e-05,
+      "loss": 1.0286,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935809156519513e-05,
+      "loss": 1.0028,
+      "step": 771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935498263653994e-05,
+      "loss": 1.0055,
+      "step": 772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935186622174723e-05,
+      "loss": 1.0679,
+      "step": 773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993487423210518e-05,
+      "loss": 1.0851,
+      "step": 774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993456109346891e-05,
+      "loss": 1.0014,
+      "step": 775
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934247206289496e-05,
+      "loss": 0.9836,
+      "step": 776
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99339325705906e-05,
+      "loss": 1.0299,
+      "step": 777
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933617186395917e-05,
+      "loss": 1.0463,
+      "step": 778
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933301053729225e-05,
+      "loss": 1.001,
+      "step": 779
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993298417261433e-05,
+      "loss": 1.0779,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932666543075113e-05,
+      "loss": 1.0371,
+      "step": 781
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932348165135507e-05,
+      "loss": 1.0389,
+      "step": 782
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932029038819507e-05,
+      "loss": 0.9969,
+      "step": 783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993170916415115e-05,
+      "loss": 1.0142,
+      "step": 784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993138854115454e-05,
+      "loss": 1.0253,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931067169853836e-05,
+      "loss": 0.9993,
+      "step": 786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993074505027325e-05,
+      "loss": 1.03,
+      "step": 787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930422182437058e-05,
+      "loss": 1.0482,
+      "step": 788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930098566369582e-05,
+      "loss": 1.0111,
+      "step": 789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992977420209521e-05,
+      "loss": 1.0297,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929449089638376e-05,
+      "loss": 1.0679,
+      "step": 791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929123229023582e-05,
+      "loss": 1.0229,
+      "step": 792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928796620275377e-05,
+      "loss": 1.0328,
+      "step": 793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928469263418376e-05,
+      "loss": 1.0156,
+      "step": 794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928141158477237e-05,
+      "loss": 0.9807,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927812305476685e-05,
+      "loss": 1.1106,
+      "step": 796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927482704441503e-05,
+      "loss": 1.0423,
+      "step": 797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927152355396515e-05,
+      "loss": 0.9698,
+      "step": 798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926821258366622e-05,
+      "loss": 0.9815,
+      "step": 799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926489413376767e-05,
+      "loss": 1.0281,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992615682045195e-05,
+      "loss": 1.0184,
+      "step": 801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925823479617242e-05,
+      "loss": 1.0114,
+      "step": 802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925489390897744e-05,
+      "loss": 1.0628,
+      "step": 803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925154554318643e-05,
+      "loss": 1.0238,
+      "step": 804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992481896990516e-05,
+      "loss": 1.0603,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924482637682584e-05,
+      "loss": 1.0468,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924145557676255e-05,
+      "loss": 1.0167,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923807729911567e-05,
+      "loss": 1.0385,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923469154413983e-05,
+      "loss": 1.0609,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923129831209008e-05,
+      "loss": 1.0253,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922789760322213e-05,
+      "loss": 1.0239,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922448941779216e-05,
+      "loss": 1.047,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99221073756057e-05,
+      "loss": 1.0111,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921765061827405e-05,
+      "loss": 0.5475,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921422000470116e-05,
+      "loss": 1.0487,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921078191559684e-05,
+      "loss": 1.0429,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992073363512202e-05,
+      "loss": 1.093,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920388331183076e-05,
+      "loss": 1.0489,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920042279768878e-05,
+      "loss": 1.0474,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991969548090549e-05,
+      "loss": 1.0458,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919347934619057e-05,
+      "loss": 0.9785,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918999640935754e-05,
+      "loss": 1.0501,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918650599881828e-05,
+      "loss": 1.0096,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918300811483576e-05,
+      "loss": 0.9792,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991795027576736e-05,
+      "loss": 1.0114,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917598992759587e-05,
+      "loss": 1.0051,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991724696248672e-05,
+      "loss": 0.9995,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916894184975298e-05,
+      "loss": 1.0508,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916540660251887e-05,
+      "loss": 0.9941,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991618638834313e-05,
+      "loss": 0.9574,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915831369275724e-05,
+      "loss": 1.0083,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915475603076414e-05,
+      "loss": 0.9835,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915119089772008e-05,
+      "loss": 1.0386,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991476182938937e-05,
+      "loss": 1.0451,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914403821955414e-05,
+      "loss": 1.0476,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914045067497117e-05,
+      "loss": 1.0025,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913685566041515e-05,
+      "loss": 0.9761,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913325317615684e-05,
+      "loss": 1.0915,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912964322246782e-05,
+      "loss": 1.0174,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912602579962e-05,
+      "loss": 1.0199,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912240090788595e-05,
+      "loss": 1.0099,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911876854753882e-05,
+      "loss": 1.0148,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991151287188523e-05,
+      "loss": 1.0299,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991114814221006e-05,
+      "loss": 0.9914,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910782665755862e-05,
+      "loss": 1.0374,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910416442550163e-05,
+      "loss": 1.0045,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910049472620564e-05,
+      "loss": 1.0098,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990968175599471e-05,
+      "loss": 1.0398,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909313292700317e-05,
+      "loss": 0.9788,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990894408276514e-05,
+      "loss": 1.0246,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908574126216996e-05,
+      "loss": 1.0593,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908203423083767e-05,
+      "loss": 1.0589,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907831973393377e-05,
+      "loss": 1.028,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907459777173827e-05,
+      "loss": 1.0248,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907086834453146e-05,
+      "loss": 0.9791,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990671314525944e-05,
+      "loss": 1.0033,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906338709620866e-05,
+      "loss": 0.9721,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905963527565636e-05,
+      "loss": 0.5433,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905587599122022e-05,
+      "loss": 1.0488,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990521092431835e-05,
+      "loss": 1.0574,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904833503182992e-05,
+      "loss": 1.0246,
+      "step": 861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904455335744395e-05,
+      "loss": 1.0602,
+      "step": 862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904076422031048e-05,
+      "loss": 0.9603,
+      "step": 863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903696762071504e-05,
+      "loss": 1.0255,
+      "step": 864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990331635589437e-05,
+      "loss": 1.0468,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902935203528307e-05,
+      "loss": 1.0463,
+      "step": 866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902553305002032e-05,
+      "loss": 1.0379,
+      "step": 867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902170660344323e-05,
+      "loss": 0.9861,
+      "step": 868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901787269584008e-05,
+      "loss": 1.0351,
+      "step": 869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990140313274998e-05,
+      "loss": 0.9802,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990101824987118e-05,
+      "loss": 1.0417,
+      "step": 871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900632620976603e-05,
+      "loss": 1.0262,
+      "step": 872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900246246095315e-05,
+      "loss": 1.0252,
+      "step": 873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899859125256417e-05,
+      "loss": 1.0149,
+      "step": 874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899471258489087e-05,
+      "loss": 1.0221,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899082645822545e-05,
+      "loss": 1.0577,
+      "step": 876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989869328728607e-05,
+      "loss": 0.9745,
+      "step": 877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989830318290901e-05,
+      "loss": 1.024,
+      "step": 878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897912332720743e-05,
+      "loss": 0.9968,
+      "step": 879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989752073675073e-05,
+      "loss": 1.0228,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989712839502847e-05,
+      "loss": 1.0632,
+      "step": 881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896735307583533e-05,
+      "loss": 1.0357,
+      "step": 882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896341474445526e-05,
+      "loss": 1.0013,
+      "step": 883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989594689564413e-05,
+      "loss": 1.0493,
+      "step": 884
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9895551571209078e-05,
+      "loss": 1.0499,
+      "step": 885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9895155501170153e-05,
+      "loss": 0.9342,
+      "step": 886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989475868555719e-05,
+      "loss": 0.9991,
+      "step": 887
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894361124400104e-05,
+      "loss": 1.0327,
+      "step": 888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893962817728842e-05,
+      "loss": 1.0211,
+      "step": 889
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893563765573414e-05,
+      "loss": 1.0425,
+      "step": 890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989316396796389e-05,
+      "loss": 1.0224,
+      "step": 891
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989276342493039e-05,
+      "loss": 0.9579,
+      "step": 892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892362136503102e-05,
+      "loss": 1.0518,
+      "step": 893
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989196010271225e-05,
+      "loss": 1.058,
+      "step": 894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989155732358814e-05,
+      "loss": 1.0556,
+      "step": 895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989115379916111e-05,
+      "loss": 1.0568,
+      "step": 896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989074952946157e-05,
+      "loss": 0.9964,
+      "step": 897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890344514519974e-05,
+      "loss": 1.0236,
+      "step": 898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889938754366843e-05,
+      "loss": 1.0069,
+      "step": 899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889532249032756e-05,
+      "loss": 1.031,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889124998548332e-05,
+      "loss": 0.9816,
+      "step": 901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988871700294426e-05,
+      "loss": 1.023,
+      "step": 902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 1.0091,
+      "step": 903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887898776500203e-05,
+      "loss": 0.5949,
+      "step": 904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887488545721864e-05,
+      "loss": 0.9978,
+      "step": 905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988707756994718e-05,
+      "loss": 1.0286,
+      "step": 906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886665849207116e-05,
+      "loss": 1.0747,
+      "step": 907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886253383532698e-05,
+      "loss": 1.0318,
+      "step": 908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885840172954998e-05,
+      "loss": 1.0479,
+      "step": 909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885426217505154e-05,
+      "loss": 1.0235,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885011517214357e-05,
+      "loss": 1.0407,
+      "step": 911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884596072113856e-05,
+      "loss": 1.0886,
+      "step": 912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884179882234946e-05,
+      "loss": 1.0164,
+      "step": 913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883762947608993e-05,
+      "loss": 1.0392,
+      "step": 914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883345268267406e-05,
+      "loss": 1.0041,
+      "step": 915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882926844241662e-05,
+      "loss": 0.9829,
+      "step": 916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882507675563284e-05,
+      "loss": 1.009,
+      "step": 917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882087762263857e-05,
+      "loss": 1.0359,
+      "step": 918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881667104375018e-05,
+      "loss": 1.0627,
+      "step": 919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881245701928467e-05,
+      "loss": 0.982,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880823554955954e-05,
+      "loss": 1.0027,
+      "step": 921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988040066348928e-05,
+      "loss": 0.9959,
+      "step": 922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879977027560317e-05,
+      "loss": 1.0244,
+      "step": 923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879552647200982e-05,
+      "loss": 1.0512,
+      "step": 924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879127522443255e-05,
+      "loss": 1.0167,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878701653319164e-05,
+      "loss": 1.0268,
+      "step": 926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878275039860795e-05,
+      "loss": 1.0501,
+      "step": 927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877847682100294e-05,
+      "loss": 0.9778,
+      "step": 928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877419580069862e-05,
+      "loss": 0.5061,
+      "step": 929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876990733801755e-05,
+      "loss": 1.0219,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876561143328287e-05,
+      "loss": 1.0184,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876130808681826e-05,
+      "loss": 0.9815,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875699729894797e-05,
+      "loss": 0.9943,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875267906999674e-05,
+      "loss": 1.0505,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874835340029004e-05,
+      "loss": 0.9969,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874402029015374e-05,
+      "loss": 1.0411,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873967973991432e-05,
+      "loss": 1.0704,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873533174989884e-05,
+      "loss": 1.0367,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873097632043493e-05,
+      "loss": 1.0093,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872661345185076e-05,
+      "loss": 0.628,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872224314447502e-05,
+      "loss": 0.9829,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871786539863706e-05,
+      "loss": 1.0134,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871348021466673e-05,
+      "loss": 1.0635,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870908759289436e-05,
+      "loss": 0.992,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870468753365098e-05,
+      "loss": 1.0198,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987002800372681e-05,
+      "loss": 0.9933,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986958651040779e-05,
+      "loss": 1.0698,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986914427344129e-05,
+      "loss": 1.0446,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986870129286064e-05,
+      "loss": 1.0373,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868257568699215e-05,
+      "loss": 1.0435,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986781310099045e-05,
+      "loss": 0.9966,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986736788976783e-05,
+      "loss": 0.9838,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866921935064907e-05,
+      "loss": 0.5514,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986647523691528e-05,
+      "loss": 0.9973,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866027795352604e-05,
+      "loss": 1.041,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865579610410595e-05,
+      "loss": 1.0209,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986513068212302e-05,
+      "loss": 1.0367,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986468101052371e-05,
+      "loss": 0.9983,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864230595646542e-05,
+      "loss": 1.0586,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863779437525454e-05,
+      "loss": 1.0577,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863327536194438e-05,
+      "loss": 0.998,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986287489168755e-05,
+      "loss": 0.972,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862421504038887e-05,
+      "loss": 1.0258,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986196737328262e-05,
+      "loss": 1.0154,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861512499452956e-05,
+      "loss": 1.0243,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986105688258418e-05,
+      "loss": 1.0052,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986060052271061e-05,
+      "loss": 0.519,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986014341986664e-05,
+      "loss": 1.0492,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985968557408671e-05,
+      "loss": 1.0056,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859226985405312e-05,
+      "loss": 1.0318,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858767653857004e-05,
+      "loss": 1.0336,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98583075794764e-05,
+      "loss": 1.0214,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857846762298157e-05,
+      "loss": 1.0109,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857385202356997e-05,
+      "loss": 1.0399,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856922899687703e-05,
+      "loss": 1.0328,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856459854325108e-05,
+      "loss": 1.0299,
+      "step": 976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855996066304096e-05,
+      "loss": 1.0153,
+      "step": 977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855531535659612e-05,
+      "loss": 1.0329,
+      "step": 978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855066262426663e-05,
+      "loss": 1.0234,
+      "step": 979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854600246640305e-05,
+      "loss": 1.024,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854133488335647e-05,
+      "loss": 1.0376,
+      "step": 981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985366598754786e-05,
+      "loss": 0.992,
+      "step": 982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985319774431217e-05,
+      "loss": 1.0296,
+      "step": 983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852728758663857e-05,
+      "loss": 1.0422,
+      "step": 984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985225903063826e-05,
+      "loss": 1.0218,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851788560270765e-05,
+      "loss": 1.0684,
+      "step": 986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851317347596827e-05,
+      "loss": 0.9935,
+      "step": 987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985084539265195e-05,
+      "loss": 1.0382,
+      "step": 988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850372695471694e-05,
+      "loss": 1.037,
+      "step": 989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849899256091672e-05,
+      "loss": 1.0009,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849425074547565e-05,
+      "loss": 0.9961,
+      "step": 991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848950150875093e-05,
+      "loss": 1.0215,
+      "step": 992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848474485110043e-05,
+      "loss": 1.0251,
+      "step": 993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847998077288255e-05,
+      "loss": 1.0507,
+      "step": 994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847520927445628e-05,
+      "loss": 1.0077,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847043035618107e-05,
+      "loss": 0.5883,
+      "step": 996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984656440184171e-05,
+      "loss": 0.9797,
+      "step": 997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984608502615249e-05,
+      "loss": 1.0037,
+      "step": 998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984560490858657e-05,
+      "loss": 0.9724,
+      "step": 999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845124049180132e-05,
+      "loss": 0.9881,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98446424479694e-05,
+      "loss": 1.0507,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844160104990665e-05,
+      "loss": 1.0179,
+      "step": 1002
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984367702028027e-05,
+      "loss": 1.0295,
+      "step": 1003
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9843193193874614e-05,
+      "loss": 0.9787,
+      "step": 1004
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984270862581015e-05,
+      "loss": 1.0232,
+      "step": 1005
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9842223316123393e-05,
+      "loss": 1.1184,
+      "step": 1006
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984173726485091e-05,
+      "loss": 1.0353,
+      "step": 1007
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841250472029316e-05,
+      "loss": 1.0157,
+      "step": 1008
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840762937695296e-05,
+      "loss": 1.0352,
+      "step": 1009
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840274661885585e-05,
+      "loss": 1.0064,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839785644636967e-05,
+      "loss": 1.0144,
+      "step": 1011
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98392958859863e-05,
+      "loss": 1.0365,
+      "step": 1012
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838805385970472e-05,
+      "loss": 0.516,
+      "step": 1013
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838314144626452e-05,
+      "loss": 1.0349,
+      "step": 1014
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837822161991248e-05,
+      "loss": 1.0202,
+      "step": 1015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837329438101927e-05,
+      "loss": 0.9971,
+      "step": 1016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836835972995622e-05,
+      "loss": 0.9919,
+      "step": 1017
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983634176670951e-05,
+      "loss": 0.979,
+      "step": 1018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983584681928083e-05,
+      "loss": 1.019,
+      "step": 1019
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835351130746873e-05,
+      "loss": 0.9473,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834854701144986e-05,
+      "loss": 0.9226,
+      "step": 1021
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983435753051258e-05,
+      "loss": 0.9856,
+      "step": 1022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983385961888711e-05,
+      "loss": 0.9799,
+      "step": 1023
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833360966306095e-05,
+      "loss": 1.0105,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832861572807106e-05,
+      "loss": 1.017,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983236143842777e-05,
+      "loss": 0.9784,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831860563205776e-05,
+      "loss": 1.0414,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831358947178855e-05,
+      "loss": 0.9861,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830856590384813e-05,
+      "loss": 1.0442,
+      "step": 1029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830353492861493e-05,
+      "loss": 1.0299,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98298496546468e-05,
+      "loss": 1.0602,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829345075778705e-05,
+      "loss": 0.9569,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828839756295223e-05,
+      "loss": 1.0088,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982833369623443e-05,
+      "loss": 1.0432,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827826895634452e-05,
+      "loss": 1.0075,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982731935453348e-05,
+      "loss": 1.0016,
+      "step": 1036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826811072969755e-05,
+      "loss": 0.9855,
+      "step": 1037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982630205098157e-05,
+      "loss": 1.0888,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825792288607284e-05,
+      "loss": 0.9887,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825281785885303e-05,
+      "loss": 0.5482,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824770542854093e-05,
+      "loss": 1.0504,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824258559552182e-05,
+      "loss": 1.0082,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823745836018132e-05,
+      "loss": 0.9863,
+      "step": 1043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982323237229059e-05,
+      "loss": 0.9912,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982271816840823e-05,
+      "loss": 1.009,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822203224409805e-05,
+      "loss": 1.0045,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821687540334116e-05,
+      "loss": 0.5888,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982117111622001e-05,
+      "loss": 1.0234,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982065395210641e-05,
+      "loss": 1.0339,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820136048032276e-05,
+      "loss": 1.0493,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981961740403663e-05,
+      "loss": 1.0465,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981909802015855e-05,
+      "loss": 1.0562,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981857789643717e-05,
+      "loss": 0.9823,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818057032911687e-05,
+      "loss": 1.0253,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981753542962134e-05,
+      "loss": 0.5516,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981701308660543e-05,
+      "loss": 1.0558,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816490003903316e-05,
+      "loss": 1.0275,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815966181554412e-05,
+      "loss": 0.9793,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981544161959819e-05,
+      "loss": 1.0274,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814916318074163e-05,
+      "loss": 1.0382,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814390277021923e-05,
+      "loss": 1.0405,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813863496481097e-05,
+      "loss": 1.0093,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813335976491387e-05,
+      "loss": 0.9712,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812807717092527e-05,
+      "loss": 1.0475,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812278718324332e-05,
+      "loss": 1.0039,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981174898022665e-05,
+      "loss": 0.9902,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811218502839405e-05,
+      "loss": 1.0103,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810687286202564e-05,
+      "loss": 1.0599,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810155330356147e-05,
+      "loss": 1.0042,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809622635340245e-05,
+      "loss": 1.0512,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809089201194987e-05,
+      "loss": 1.0354,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980855502796057e-05,
+      "loss": 0.974,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980802011567724e-05,
+      "loss": 1.0245,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807484464385304e-05,
+      "loss": 0.6193,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806948074125123e-05,
+      "loss": 1.0258,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980641094493711e-05,
+      "loss": 1.0061,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805873076861734e-05,
+      "loss": 1.0109,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980533446993953e-05,
+      "loss": 0.979,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980479512421107e-05,
+      "loss": 0.9665,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804255039717e-05,
+      "loss": 0.9975,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803714216498013e-05,
+      "loss": 1.074,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803172654594855e-05,
+      "loss": 1.021,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802630354048335e-05,
+      "loss": 1.0308,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980208731489931e-05,
+      "loss": 1.0299,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98015435371887e-05,
+      "loss": 1.0218,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800999020957475e-05,
+      "loss": 1.0171,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800453766246668e-05,
+      "loss": 1.0133,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799907773097355e-05,
+      "loss": 1.081,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799361041550678e-05,
+      "loss": 1.0556,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798813571647835e-05,
+      "loss": 1.0024,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798265363430075e-05,
+      "loss": 1.0301,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797716416938703e-05,
+      "loss": 1.0229,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797166732215078e-05,
+      "loss": 1.0078,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979661630930062e-05,
+      "loss": 0.9732,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796065148236803e-05,
+      "loss": 1.0109,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795513249065155e-05,
+      "loss": 1.0337,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979496061182726e-05,
+      "loss": 1.0482,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794407236564753e-05,
+      "loss": 1.0004,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979385312331934e-05,
+      "loss": 0.9799,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979329827213276e-05,
+      "loss": 1.0056,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979274268304683e-05,
+      "loss": 1.0234,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792186356103403e-05,
+      "loss": 1.0372,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791629291344404e-05,
+      "loss": 0.9961,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791071488811802e-05,
+      "loss": 0.983,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790512948547633e-05,
+      "loss": 1.0488,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789953670593972e-05,
+      "loss": 1.0067,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978939365499296e-05,
+      "loss": 1.0117,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97888329017868e-05,
+      "loss": 1.0252,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978827141101774e-05,
+      "loss": 1.0533,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978770918272808e-05,
+      "loss": 1.0501,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787146216960196e-05,
+      "loss": 1.0545,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786582513756496e-05,
+      "loss": 1.0121,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786018073159458e-05,
+      "loss": 1.0752,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785452895211606e-05,
+      "loss": 0.9874,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784886979955526e-05,
+      "loss": 1.0612,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784320327433865e-05,
+      "loss": 0.994,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783752937689312e-05,
+      "loss": 0.6095,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978318481076462e-05,
+      "loss": 1.0447,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782615946702596e-05,
+      "loss": 1.0117,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782046345546102e-05,
+      "loss": 1.013,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 1.0335,
+      "step": 1121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9780904932121435e-05,
+      "loss": 1.0534,
+      "step": 1122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9780333119939264e-05,
+      "loss": 1.0034,
+      "step": 1123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779760570834626e-05,
+      "loss": 0.9978,
+      "step": 1124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779187284850665e-05,
+      "loss": 0.9905,
+      "step": 1125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778613262030577e-05,
+      "loss": 1.0278,
+      "step": 1126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977803850241761e-05,
+      "loss": 0.9975,
+      "step": 1127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977746300605507e-05,
+      "loss": 1.0271,
+      "step": 1128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776886772986325e-05,
+      "loss": 0.981,
+      "step": 1129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977630980325479e-05,
+      "loss": 1.0023,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977573209690393e-05,
+      "loss": 1.004,
+      "step": 1131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775153653977284e-05,
+      "loss": 0.9869,
+      "step": 1132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774574474518432e-05,
+      "loss": 1.0247,
+      "step": 1133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773994558571016e-05,
+      "loss": 0.97,
+      "step": 1134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977341390617873e-05,
+      "loss": 0.965,
+      "step": 1135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977283251738532e-05,
+      "loss": 1.0336,
+      "step": 1136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97722503922346e-05,
+      "loss": 0.9681,
+      "step": 1137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771667530770427e-05,
+      "loss": 1.0338,
+      "step": 1138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977108393303672e-05,
+      "loss": 1.0003,
+      "step": 1139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977049959907745e-05,
+      "loss": 1.0285,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769914528936646e-05,
+      "loss": 0.9672,
+      "step": 1141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976932872265839e-05,
+      "loss": 1.0507,
+      "step": 1142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976874218028682e-05,
+      "loss": 0.5818,
+      "step": 1143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768154901866136e-05,
+      "loss": 1.0499,
+      "step": 1144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976756688744058e-05,
+      "loss": 1.0455,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766978137054466e-05,
+      "loss": 1.0082,
+      "step": 1146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766388650752152e-05,
+      "loss": 1.021,
+      "step": 1147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765798428578045e-05,
+      "loss": 1.0006,
+      "step": 1148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765207470576632e-05,
+      "loss": 0.9496,
+      "step": 1149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976461577679243e-05,
+      "loss": 1.0152,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764023347270026e-05,
+      "loss": 1.0095,
+      "step": 1151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976343018205405e-05,
+      "loss": 0.9566,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762836281189207e-05,
+      "loss": 0.9988,
+      "step": 1153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976224164472024e-05,
+      "loss": 1.0282,
+      "step": 1154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761646272691954e-05,
+      "loss": 0.9785,
+      "step": 1155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761050165149208e-05,
+      "loss": 1.0181,
+      "step": 1156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760453322136916e-05,
+      "loss": 0.9921,
+      "step": 1157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975985574370005e-05,
+      "loss": 1.037,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975925742988364e-05,
+      "loss": 1.0361,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758658380732758e-05,
+      "loss": 0.9927,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758058596292548e-05,
+      "loss": 0.9751,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757458076608204e-05,
+      "loss": 0.97,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975685682172497e-05,
+      "loss": 1.0311,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756254831688142e-05,
+      "loss": 1.0221,
+      "step": 1164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755652106543094e-05,
+      "loss": 1.0135,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755048646335227e-05,
+      "loss": 0.9861,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754444451110014e-05,
+      "loss": 1.0043,
+      "step": 1167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753839520912984e-05,
+      "loss": 1.0402,
+      "step": 1168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753233855789705e-05,
+      "loss": 0.9757,
+      "step": 1169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752627455785825e-05,
+      "loss": 1.0163,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975202032094703e-05,
+      "loss": 0.9754,
+      "step": 1171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751412451319068e-05,
+      "loss": 0.9919,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975080384694773e-05,
+      "loss": 0.9881,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975019450787889e-05,
+      "loss": 0.9752,
+      "step": 1174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749584434158448e-05,
+      "loss": 1.0003,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748973625832373e-05,
+      "loss": 0.9887,
+      "step": 1176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974836208294669e-05,
+      "loss": 1.0362,
+      "step": 1177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747749805547474e-05,
+      "loss": 1.0306,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747136793680864e-05,
+      "loss": 1.0011,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746523047393046e-05,
+      "loss": 0.9992,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745908566730262e-05,
+      "loss": 0.9719,
+      "step": 1181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745293351738813e-05,
+      "loss": 1.0404,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744677402465053e-05,
+      "loss": 1.0282,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744060718955396e-05,
+      "loss": 0.9624,
+      "step": 1184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743443301256303e-05,
+      "loss": 1.0397,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97428251494143e-05,
+      "loss": 0.9596,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742206263475956e-05,
+      "loss": 0.9868,
+      "step": 1187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974158664348791e-05,
+      "loss": 1.002,
+      "step": 1188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740966289496844e-05,
+      "loss": 1.0262,
+      "step": 1189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97403452015495e-05,
+      "loss": 0.9986,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739723379692676e-05,
+      "loss": 1.0014,
+      "step": 1191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739100823973226e-05,
+      "loss": 1.0144,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738477534438055e-05,
+      "loss": 1.0404,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973785351113413e-05,
+      "loss": 0.9778,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737228754108467e-05,
+      "loss": 0.9798,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973660326340814e-05,
+      "loss": 1.0603,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973597703908028e-05,
+      "loss": 0.9713,
+      "step": 1197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973535008117207e-05,
+      "loss": 0.9862,
+      "step": 1198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734722389730748e-05,
+      "loss": 1.0425,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973409396480361e-05,
+      "loss": 1.068,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733464806438007e-05,
+      "loss": 0.9697,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732834914681347e-05,
+      "loss": 1.0064,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732204289581088e-05,
+      "loss": 0.9864,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973157293118474e-05,
+      "loss": 1.0662,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730940839539884e-05,
+      "loss": 0.9463,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730308014694143e-05,
+      "loss": 1.0029,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97296744566952e-05,
+      "loss": 0.9912,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729040165590784e-05,
+      "loss": 0.9851,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728405141428698e-05,
+      "loss": 0.9708,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727769384256784e-05,
+      "loss": 0.9995,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972713289412294e-05,
+      "loss": 1.0301,
+      "step": 1211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726495671075135e-05,
+      "loss": 1.0308,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725857715161375e-05,
+      "loss": 0.9978,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972521902642973e-05,
+      "loss": 1.0431,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972457960492832e-05,
+      "loss": 1.003,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972393945070533e-05,
+      "loss": 1.0094,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972329856380899e-05,
+      "loss": 0.9913,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972265694428759e-05,
+      "loss": 1.0597,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722014592189472e-05,
+      "loss": 0.9955,
+      "step": 1219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972137150756304e-05,
+      "loss": 1.0057,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720727690456745e-05,
+      "loss": 0.5605,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720083140919097e-05,
+      "loss": 1.0589,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719437858998667e-05,
+      "loss": 1.0405,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718791844744068e-05,
+      "loss": 1.0264,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718145098203977e-05,
+      "loss": 0.9484,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971749761942713e-05,
+      "loss": 0.9986,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716849408462308e-05,
+      "loss": 1.0054,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716200465358352e-05,
+      "loss": 0.9999,
+      "step": 1228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715550790164156e-05,
+      "loss": 1.0115,
+      "step": 1229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714900382928674e-05,
+      "loss": 1.0319,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714249243700916e-05,
+      "loss": 1.0032,
+      "step": 1231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971359737252994e-05,
+      "loss": 1.0399,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712944769464864e-05,
+      "loss": 1.0343,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712291434554858e-05,
+      "loss": 1.0598,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971163736784915e-05,
+      "loss": 1.0336,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971098256939702e-05,
+      "loss": 1.0086,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710327039247814e-05,
+      "loss": 1.0375,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709670777450912e-05,
+      "loss": 1.0092,
+      "step": 1238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9709013784055773e-05,
+      "loss": 1.0496,
+      "step": 1239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970835605911189e-05,
+      "loss": 1.0358,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9707697602668827e-05,
+      "loss": 0.9372,
+      "step": 1241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9707038414776195e-05,
+      "loss": 1.0137,
+      "step": 1242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706378495483664e-05,
+      "loss": 0.9756,
+      "step": 1243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705717844840955e-05,
+      "loss": 1.0137,
+      "step": 1244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705056462897846e-05,
+      "loss": 1.0472,
+      "step": 1245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704394349704174e-05,
+      "loss": 1.0384,
+      "step": 1246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970373150530982e-05,
+      "loss": 1.0493,
+      "step": 1247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703067929764736e-05,
+      "loss": 0.9746,
+      "step": 1248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702403623118918e-05,
+      "loss": 0.9784,
+      "step": 1249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701738585422416e-05,
+      "loss": 1.0078,
+      "step": 1250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701072816725343e-05,
+      "loss": 0.4621,
+      "step": 1251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970040631707786e-05,
+      "loss": 1.0087,
+      "step": 1252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969973908653019e-05,
+      "loss": 1.029,
+      "step": 1253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699071125132604e-05,
+      "loss": 1.017,
+      "step": 1254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698402432935432e-05,
+      "loss": 1.0194,
+      "step": 1255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697733009989056e-05,
+      "loss": 0.9775,
+      "step": 1256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969706285634392e-05,
+      "loss": 1.0116,
+      "step": 1257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696391972050516e-05,
+      "loss": 1.0262,
+      "step": 1258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969572035715939e-05,
+      "loss": 1.0511,
+      "step": 1259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695048011721152e-05,
+      "loss": 0.9804,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694374935786457e-05,
+      "loss": 0.9627,
+      "step": 1261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969370112940602e-05,
+      "loss": 0.9886,
+      "step": 1262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693026592630614e-05,
+      "loss": 1.0348,
+      "step": 1263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692351325511066e-05,
+      "loss": 0.9972,
+      "step": 1264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691675328098244e-05,
+      "loss": 0.9746,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690998600443096e-05,
+      "loss": 1.0768,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690321142596602e-05,
+      "loss": 1.0384,
+      "step": 1267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689642954609808e-05,
+      "loss": 1.0245,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688964036533817e-05,
+      "loss": 0.9626,
+      "step": 1269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688284388419784e-05,
+      "loss": 1.0153,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687604010318916e-05,
+      "loss": 0.9672,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968692290228248e-05,
+      "loss": 0.97,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686241064361792e-05,
+      "loss": 1.0112,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685558496608236e-05,
+      "loss": 1.0313,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684875199073227e-05,
+      "loss": 1.0086,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684191171808262e-05,
+      "loss": 0.984,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683506414864878e-05,
+      "loss": 0.9909,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682820928294664e-05,
+      "loss": 1.0015,
+      "step": 1278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968213471214927e-05,
+      "loss": 0.9083,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968144776648041e-05,
+      "loss": 0.9187,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968076009133984e-05,
+      "loss": 1.0291,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680071686779368e-05,
+      "loss": 1.0045,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679382552850867e-05,
+      "loss": 1.0001,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678692689606267e-05,
+      "loss": 1.0423,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678002097097537e-05,
+      "loss": 1.0136,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967731077537672e-05,
+      "loss": 0.9853,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676618724495903e-05,
+      "loss": 1.0298,
+      "step": 1287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675925944507226e-05,
+      "loss": 1.0012,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675232435462893e-05,
+      "loss": 1.0095,
+      "step": 1289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674538197415157e-05,
+      "loss": 0.972,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967384323041633e-05,
+      "loss": 1.0384,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673147534518767e-05,
+      "loss": 1.0249,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967245110977489e-05,
+      "loss": 1.0298,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671753956237187e-05,
+      "loss": 1.0066,
+      "step": 1294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671056073958165e-05,
+      "loss": 0.935,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670357462990418e-05,
+      "loss": 0.9983,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966965812338659e-05,
+      "loss": 1.0166,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668958055199365e-05,
+      "loss": 1.0047,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966825725848149e-05,
+      "loss": 1.0234,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966755573328578e-05,
+      "loss": 1.0422,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666853479665083e-05,
+      "loss": 1.0218,
+      "step": 1301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666150497672312e-05,
+      "loss": 1.0273,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665446787360444e-05,
+      "loss": 1.0776,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9664742348782492e-05,
+      "loss": 0.6006,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966403718199154e-05,
+      "loss": 1.0174,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663331287040713e-05,
+      "loss": 1.0554,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662624663983204e-05,
+      "loss": 1.0067,
+      "step": 1307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661917312872258e-05,
+      "loss": 1.011,
+      "step": 1308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661209233761167e-05,
+      "loss": 0.991,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660500426703282e-05,
+      "loss": 0.9441,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965979089175201e-05,
+      "loss": 0.9952,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965908062896082e-05,
+      "loss": 1.0565,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965836963838322e-05,
+      "loss": 1.0215,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657657920072784e-05,
+      "loss": 0.9845,
+      "step": 1314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965694547408314e-05,
+      "loss": 1.0435,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656232300467964e-05,
+      "loss": 0.9678,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655518399281e-05,
+      "loss": 0.9986,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965480377057603e-05,
+      "loss": 0.9729,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96540884144069e-05,
+      "loss": 1.0311,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965337233082752e-05,
+      "loss": 0.9986,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965265551989183e-05,
+      "loss": 0.9861,
+      "step": 1321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651937981653854e-05,
+      "loss": 1.0051,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965121971616765e-05,
+      "loss": 0.9987,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650500723487335e-05,
+      "loss": 1.0236,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964978100366709e-05,
+      "loss": 0.9767,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964906055676114e-05,
+      "loss": 1.0246,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964833938282377e-05,
+      "loss": 1.0074,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647617481909314e-05,
+      "loss": 1.0026,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964689485407217e-05,
+      "loss": 1.0308,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964617149936679e-05,
+      "loss": 0.9654,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645447417847663e-05,
+      "loss": 0.9917,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644722609569363e-05,
+      "loss": 1.0244,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964399707458649e-05,
+      "loss": 1.0019,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643270812953722e-05,
+      "loss": 0.9863,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964254382472577e-05,
+      "loss": 0.9504,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641816109957415e-05,
+      "loss": 1.0312,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641087668703494e-05,
+      "loss": 0.9741,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640358501018885e-05,
+      "loss": 0.9982,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639628606958535e-05,
+      "loss": 1.0621,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638897986577435e-05,
+      "loss": 1.0068,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963816663993064e-05,
+      "loss": 1.0095,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637434567073246e-05,
+      "loss": 0.9645,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963670176806042e-05,
+      "loss": 1.0045,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635968242947378e-05,
+      "loss": 0.9916,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963523399178939e-05,
+      "loss": 1.0148,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963449901464177e-05,
+      "loss": 0.926,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633763311559904e-05,
+      "loss": 1.017,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633026882599228e-05,
+      "loss": 1.0322,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632289727815227e-05,
+      "loss": 1.0249,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631551847263438e-05,
+      "loss": 1.0074,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630813240999468e-05,
+      "loss": 1.0454,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963007390907896e-05,
+      "loss": 1.0417,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962933385155763e-05,
+      "loss": 1.0011,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962859306849123e-05,
+      "loss": 1.0059,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627851559935583e-05,
+      "loss": 1.0053,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627109325946557e-05,
+      "loss": 1.0094,
+      "step": 1356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962636636658008e-05,
+      "loss": 1.023,
+      "step": 1357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962562268189213e-05,
+      "loss": 0.5846,
+      "step": 1358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9624878271938737e-05,
+      "loss": 1.0143,
+      "step": 1359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9624133136775998e-05,
+      "loss": 0.96,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962338727646005e-05,
+      "loss": 0.986,
+      "step": 1361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622640691047104e-05,
+      "loss": 0.9963,
+      "step": 1362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621893380593398e-05,
+      "loss": 0.9915,
+      "step": 1363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621145345155247e-05,
+      "loss": 0.9804,
+      "step": 1364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620396584789018e-05,
+      "loss": 1.0466,
+      "step": 1365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619647099551118e-05,
+      "loss": 0.9835,
+      "step": 1366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618896889498024e-05,
+      "loss": 1.0197,
+      "step": 1367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618145954686265e-05,
+      "loss": 1.0086,
+      "step": 1368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617394295172415e-05,
+      "loss": 0.9895,
+      "step": 1369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961664191101312e-05,
+      "loss": 0.9703,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961588880226506e-05,
+      "loss": 0.9797,
+      "step": 1371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615134968984984e-05,
+      "loss": 0.992,
+      "step": 1372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614380411229693e-05,
+      "loss": 0.9833,
+      "step": 1373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613625129056037e-05,
+      "loss": 1.0031,
+      "step": 1374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961286912252093e-05,
+      "loss": 1.0139,
+      "step": 1375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612112391681327e-05,
+      "loss": 0.9507,
+      "step": 1376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611354936594255e-05,
+      "loss": 1.0264,
+      "step": 1377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961059675731678e-05,
+      "loss": 0.9953,
+      "step": 1378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609837853906028e-05,
+      "loss": 0.994,
+      "step": 1379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609078226419186e-05,
+      "loss": 0.9956,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608317874913484e-05,
+      "loss": 0.9783,
+      "step": 1381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607556799446218e-05,
+      "loss": 1.06,
+      "step": 1382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606795000074732e-05,
+      "loss": 1.0084,
+      "step": 1383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960603247685642e-05,
+      "loss": 0.9513,
+      "step": 1384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960526922984874e-05,
+      "loss": 0.96,
+      "step": 1385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604505259109207e-05,
+      "loss": 1.0103,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960374056469537e-05,
+      "loss": 0.9987,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602975146664863e-05,
+      "loss": 1.0013,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602209005075345e-05,
+      "loss": 0.9734,
+      "step": 1389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601442139984548e-05,
+      "loss": 0.9387,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600674551450252e-05,
+      "loss": 1.0003,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599906239530294e-05,
+      "loss": 1.0182,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599137204282566e-05,
+      "loss": 1.0398,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959836744576501e-05,
+      "loss": 1.0293,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597596964035627e-05,
+      "loss": 0.9999,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596825759152466e-05,
+      "loss": 0.9825,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596053831173643e-05,
+      "loss": 1.017,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595281180157315e-05,
+      "loss": 1.0316,
+      "step": 1398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594507806161703e-05,
+      "loss": 0.9782,
+      "step": 1399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593733709245077e-05,
+      "loss": 0.9889,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959295888946576e-05,
+      "loss": 1.0444,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592183346882135e-05,
+      "loss": 0.9518,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959140708155264e-05,
+      "loss": 0.9741,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590630093535762e-05,
+      "loss": 0.9867,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958985238289005e-05,
+      "loss": 1.0124,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958907394967409e-05,
+      "loss": 1.0077,
+      "step": 1406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588294793946547e-05,
+      "loss": 1.0822,
+      "step": 1407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587514915766124e-05,
+      "loss": 1.0109,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958673431519158e-05,
+      "loss": 1.0002,
+      "step": 1409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585952992281737e-05,
+      "loss": 0.9632,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958517094709546e-05,
+      "loss": 0.987,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584388179691683e-05,
+      "loss": 0.9979,
+      "step": 1412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583604690129375e-05,
+      "loss": 0.999,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958282047846757e-05,
+      "loss": 0.9584,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582035544765366e-05,
+      "loss": 1.0148,
+      "step": 1415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581249889081898e-05,
+      "loss": 1.0279,
+      "step": 1416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580463511476365e-05,
+      "loss": 0.9184,
+      "step": 1417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957967641200802e-05,
+      "loss": 1.0017,
+      "step": 1418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578888590736163e-05,
+      "loss": 1.0485,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578100047720164e-05,
+      "loss": 1.0153,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577310783019432e-05,
+      "loss": 0.9481,
+      "step": 1421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576520796693433e-05,
+      "loss": 1.0177,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575730088801696e-05,
+      "loss": 0.963,
+      "step": 1423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574938659403798e-05,
+      "loss": 1.0057,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957414650855937e-05,
+      "loss": 0.9297,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573353636328094e-05,
+      "loss": 1.0345,
+      "step": 1426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957256004276972e-05,
+      "loss": 1.0227,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571765727944034e-05,
+      "loss": 0.9637,
+      "step": 1428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957097069191089e-05,
+      "loss": 0.981,
+      "step": 1429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570174934730195e-05,
+      "loss": 0.9984,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95693784564619e-05,
+      "loss": 1.0086,
+      "step": 1431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568581257166025e-05,
+      "loss": 1.023,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956778333690263e-05,
+      "loss": 1.028,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566984695731843e-05,
+      "loss": 1.0099,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566185333713835e-05,
+      "loss": 1.0283,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565385250908838e-05,
+      "loss": 0.9728,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564584447377136e-05,
+      "loss": 1.0092,
+      "step": 1437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563782923179063e-05,
+      "loss": 0.9603,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562980678375014e-05,
+      "loss": 0.9741,
+      "step": 1439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562177713025444e-05,
+      "loss": 1.0033,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956137402719084e-05,
+      "loss": 0.9871,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956056962093177e-05,
+      "loss": 1.03,
+      "step": 1442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559764494308838e-05,
+      "loss": 0.9806,
+      "step": 1443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955895864738271e-05,
+      "loss": 0.9806,
+      "step": 1444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558152080214102e-05,
+      "loss": 0.5907,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557344792863788e-05,
+      "loss": 0.9846,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556536785392598e-05,
+      "loss": 1.012,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 1.0104,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955491861033116e-05,
+      "loss": 1.0029,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554108442862836e-05,
+      "loss": 1.0136,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553297555517488e-05,
+      "loss": 0.5597,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552485948356206e-05,
+      "loss": 0.9958,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955167362144015e-05,
+      "loss": 1.0037,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955086057483052e-05,
+      "loss": 0.9702,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955004680858858e-05,
+      "loss": 0.999,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954923232277565e-05,
+      "loss": 1.0258,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548417117453092e-05,
+      "loss": 0.9961,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547601192682335e-05,
+      "loss": 0.9983,
+      "step": 1458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546784548524852e-05,
+      "loss": 0.9822,
+      "step": 1459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545967185042176e-05,
+      "loss": 0.9503,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545149102295896e-05,
+      "loss": 1.0249,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544330300347655e-05,
+      "loss": 1.0041,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543510779259135e-05,
+      "loss": 0.9407,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542690539092098e-05,
+      "loss": 1.0271,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541869579908343e-05,
+      "loss": 0.983,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541047901769724e-05,
+      "loss": 1.0073,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540225504738156e-05,
+      "loss": 0.9852,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539402388875598e-05,
+      "loss": 0.9558,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9538578554244082e-05,
+      "loss": 1.0219,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953775400090567e-05,
+      "loss": 1.0478,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536928728922496e-05,
+      "loss": 0.9789,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953610273835674e-05,
+      "loss": 1.003,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535276029270636e-05,
+      "loss": 0.9684,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953444860172648e-05,
+      "loss": 0.9999,
+      "step": 1474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9533620455786613e-05,
+      "loss": 1.0095,
+      "step": 1475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9532791591513437e-05,
+      "loss": 0.994,
+      "step": 1476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9531962008969396e-05,
+      "loss": 1.0331,
+      "step": 1477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9531131708217005e-05,
+      "loss": 1.0127,
+      "step": 1478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530300689318823e-05,
+      "loss": 0.5713,
+      "step": 1479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529468952337468e-05,
+      "loss": 1.0196,
+      "step": 1480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95286364973356e-05,
+      "loss": 1.0182,
+      "step": 1481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952780332437595e-05,
+      "loss": 1.0339,
+      "step": 1482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526969433521298e-05,
+      "loss": 0.9914,
+      "step": 1483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526134824834466e-05,
+      "loss": 0.9887,
+      "step": 1484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525299498378348e-05,
+      "loss": 0.9978,
+      "step": 1485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952446345421588e-05,
+      "loss": 0.983,
+      "step": 1486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523626692410052e-05,
+      "loss": 0.9597,
+      "step": 1487
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952278921302392e-05,
+      "loss": 0.95,
+      "step": 1488
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521951016120582e-05,
+      "loss": 0.9799,
+      "step": 1489
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521112101763197e-05,
+      "loss": 0.9727,
+      "step": 1490
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520272470014964e-05,
+      "loss": 0.9978,
+      "step": 1491
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951943212093916e-05,
+      "loss": 1.0033,
+      "step": 1492
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518591054599095e-05,
+      "loss": 1.0483,
+      "step": 1493
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517749271058148e-05,
+      "loss": 0.9941,
+      "step": 1494
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951690677037974e-05,
+      "loss": 0.9582,
+      "step": 1495
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516063552627347e-05,
+      "loss": 1.0139,
+      "step": 1496
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515219617864515e-05,
+      "loss": 1.0185,
+      "step": 1497
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514374966154826e-05,
+      "loss": 1.0109,
+      "step": 1498
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951352959756192e-05,
+      "loss": 1.0316,
+      "step": 1499
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512683512149494e-05,
+      "loss": 0.9725,
+      "step": 1500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511836709981306e-05,
+      "loss": 1.0353,
+      "step": 1501
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510989191121146e-05,
+      "loss": 0.5344,
+      "step": 1502
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951014095563289e-05,
+      "loss": 1.0142,
+      "step": 1503
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950929200358043e-05,
+      "loss": 0.9993,
+      "step": 1504
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950844233502775e-05,
+      "loss": 1.0073,
+      "step": 1505
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507591950038864e-05,
+      "loss": 0.9897,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506740848677845e-05,
+      "loss": 1.0502,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950588903100882e-05,
+      "loss": 0.9698,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505036497095975e-05,
+      "loss": 1.0233,
+      "step": 1509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504183247003544e-05,
+      "loss": 0.989,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503329280795816e-05,
+      "loss": 1.0191,
+      "step": 1511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950247459853714e-05,
+      "loss": 0.9895,
+      "step": 1512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950161920029191e-05,
+      "loss": 1.0255,
+      "step": 1513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500763086124576e-05,
+      "loss": 1.0105,
+      "step": 1514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949990625609965e-05,
+      "loss": 1.0191,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499048710281686e-05,
+      "loss": 1.0095,
+      "step": 1516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498190448735303e-05,
+      "loss": 0.9757,
+      "step": 1517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497331471525165e-05,
+      "loss": 1.0346,
+      "step": 1518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496471778715996e-05,
+      "loss": 1.0,
+      "step": 1519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949561137037257e-05,
+      "loss": 0.9779,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494750246559715e-05,
+      "loss": 1.0295,
+      "step": 1521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949388840734232e-05,
+      "loss": 0.9731,
+      "step": 1522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493025852785314e-05,
+      "loss": 1.0336,
+      "step": 1523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492162582953698e-05,
+      "loss": 1.0312,
+      "step": 1524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949129859791251e-05,
+      "loss": 0.9811,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490433897726852e-05,
+      "loss": 0.9859,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489568482461872e-05,
+      "loss": 1.018,
+      "step": 1527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948870235218279e-05,
+      "loss": 0.9821,
+      "step": 1528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487835506954846e-05,
+      "loss": 0.9822,
+      "step": 1529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948696794684337e-05,
+      "loss": 1.0196,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948609967191373e-05,
+      "loss": 1.0308,
+      "step": 1531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948523068223134e-05,
+      "loss": 1.0325,
+      "step": 1532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948436097786169e-05,
+      "loss": 0.9875,
+      "step": 1533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948349055887029e-05,
+      "loss": 0.995,
+      "step": 1534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482619425322738e-05,
+      "loss": 0.9771,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948174757728467e-05,
+      "loss": 1.0044,
+      "step": 1536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480875014821776e-05,
+      "loss": 1.0284,
+      "step": 1537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480001737999796e-05,
+      "loss": 0.9587,
+      "step": 1538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479127746884537e-05,
+      "loss": 1.018,
+      "step": 1539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478253041541848e-05,
+      "loss": 1.0546,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947737762203764e-05,
+      "loss": 1.007,
+      "step": 1541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476501488437864e-05,
+      "loss": 0.9582,
+      "step": 1542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475624640808542e-05,
+      "loss": 0.9833,
+      "step": 1543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474747079215745e-05,
+      "loss": 0.9895,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947386880372559e-05,
+      "loss": 0.9745,
+      "step": 1545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947298981440425e-05,
+      "loss": 1.0303,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472110111317958e-05,
+      "loss": 1.0217,
+      "step": 1547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471229694533003e-05,
+      "loss": 1.0094,
+      "step": 1548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947034856411571e-05,
+      "loss": 1.0097,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469466720132478e-05,
+      "loss": 0.9883,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946858416264975e-05,
+      "loss": 0.9891,
+      "step": 1551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467700891734027e-05,
+      "loss": 0.9927,
+      "step": 1552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466816907451857e-05,
+      "loss": 1.0364,
+      "step": 1553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946593220986985e-05,
+      "loss": 1.0176,
+      "step": 1554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465046799054657e-05,
+      "loss": 0.9606,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464160675073e-05,
+      "loss": 1.0079,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463273837991643e-05,
+      "loss": 1.0082,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946238628787741e-05,
+      "loss": 1.0326,
+      "step": 1558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946149802479717e-05,
+      "loss": 0.9947,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460609048817852e-05,
+      "loss": 0.9595,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945971936000645e-05,
+      "loss": 1.0335,
+      "step": 1561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945882895842998e-05,
+      "loss": 1.0311,
+      "step": 1562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457937844155546e-05,
+      "loss": 0.9985,
+      "step": 1563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457046017250283e-05,
+      "loss": 0.9337,
+      "step": 1564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456153477781395e-05,
+      "loss": 0.9943,
+      "step": 1565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455260225816128e-05,
+      "loss": 0.5809,
+      "step": 1566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454366261421786e-05,
+      "loss": 0.9681,
+      "step": 1567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945347158466573e-05,
+      "loss": 0.968,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452576195615366e-05,
+      "loss": 0.9789,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451680094338163e-05,
+      "loss": 1.0004,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945078328090164e-05,
+      "loss": 0.9795,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449885755373375e-05,
+      "loss": 0.6491,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448987517820982e-05,
+      "loss": 1.0279,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448088568312146e-05,
+      "loss": 0.9982,
+      "step": 1574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944718890691461e-05,
+      "loss": 0.9758,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446288533696145e-05,
+      "loss": 0.9971,
+      "step": 1576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94453874487246e-05,
+      "loss": 1.0085,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944448565206787e-05,
+      "loss": 0.5698,
+      "step": 1578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443583143793904e-05,
+      "loss": 0.9338,
+      "step": 1579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94426799239707e-05,
+      "loss": 0.9982,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441775992666313e-05,
+      "loss": 1.022,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440871349948856e-05,
+      "loss": 1.002,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943996599588649e-05,
+      "loss": 0.9998,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439059930547427e-05,
+      "loss": 0.9678,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438153153999942e-05,
+      "loss": 0.9854,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437245666312353e-05,
+      "loss": 1.0022,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436337467553042e-05,
+      "loss": 0.996,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943542855779044e-05,
+      "loss": 0.9837,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434518937093025e-05,
+      "loss": 0.9744,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943360860552934e-05,
+      "loss": 1.0199,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432697563167974e-05,
+      "loss": 1.0183,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943178581007757e-05,
+      "loss": 0.9478,
+      "step": 1592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9430873346326825e-05,
+      "loss": 0.9761,
+      "step": 1593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9429960171984496e-05,
+      "loss": 1.0024,
+      "step": 1594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9429046287119387e-05,
+      "loss": 0.9957,
+      "step": 1595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9428131691800352e-05,
+      "loss": 0.9608,
+      "step": 1596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427216386096313e-05,
+      "loss": 0.9934,
+      "step": 1597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426300370076224e-05,
+      "loss": 0.9354,
+      "step": 1598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425383643809112e-05,
+      "loss": 0.977,
+      "step": 1599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942446620736405e-05,
+      "loss": 0.9928,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942354806081016e-05,
+      "loss": 0.939,
+      "step": 1601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422629204216627e-05,
+      "loss": 0.9683,
+      "step": 1602
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421709637652683e-05,
+      "loss": 1.0356,
+      "step": 1603
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420789361187607e-05,
+      "loss": 0.9758,
+      "step": 1604
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419868374890754e-05,
+      "loss": 0.986,
+      "step": 1605
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418946678831507e-05,
+      "loss": 1.0344,
+      "step": 1606
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418024273079318e-05,
+      "loss": 0.9976,
+      "step": 1607
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417101157703688e-05,
+      "loss": 0.9582,
+      "step": 1608
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416177332774162e-05,
+      "loss": 0.9864,
+      "step": 1609
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415252798360363e-05,
+      "loss": 1.0008,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941432755453194e-05,
+      "loss": 1.077,
+      "step": 1611
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413401601358616e-05,
+      "loss": 0.9764,
+      "step": 1612
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412474938910157e-05,
+      "loss": 1.0172,
+      "step": 1613
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941154756725638e-05,
+      "loss": 0.5723,
+      "step": 1614
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410619486467165e-05,
+      "loss": 0.9835,
+      "step": 1615
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409690696612436e-05,
+      "loss": 1.0037,
+      "step": 1616
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940876119776218e-05,
+      "loss": 0.999,
+      "step": 1617
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940783098998643e-05,
+      "loss": 0.979,
+      "step": 1618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406900073355275e-05,
+      "loss": 0.9558,
+      "step": 1619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405968447938855e-05,
+      "loss": 0.9866,
+      "step": 1620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940503611380737e-05,
+      "loss": 1.0152,
+      "step": 1621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404103071031062e-05,
+      "loss": 1.043,
+      "step": 1622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940316931968024e-05,
+      "loss": 0.9495,
+      "step": 1623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402234859825257e-05,
+      "loss": 1.013,
+      "step": 1624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401299691536523e-05,
+      "loss": 0.9105,
+      "step": 1625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94003638148845e-05,
+      "loss": 1.0002,
+      "step": 1626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399427229939704e-05,
+      "loss": 0.9951,
+      "step": 1627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93984899367727e-05,
+      "loss": 1.0188,
+      "step": 1628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939755193545412e-05,
+      "loss": 0.9463,
+      "step": 1629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939661322605463e-05,
+      "loss": 0.9548,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395673808644968e-05,
+      "loss": 1.0102,
+      "step": 1631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939473368329591e-05,
+      "loss": 1.0019,
+      "step": 1632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393792850078294e-05,
+      "loss": 0.9536,
+      "step": 1633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939285130906301e-05,
+      "loss": 0.9988,
+      "step": 1634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391909060321002e-05,
+      "loss": 0.9669,
+      "step": 1635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939096610392326e-05,
+      "loss": 0.9634,
+      "step": 1636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939002243994084e-05,
+      "loss": 1.0045,
+      "step": 1637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389078068444843e-05,
+      "loss": 1.0372,
+      "step": 1638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388132989506422e-05,
+      "loss": 1.0129,
+      "step": 1639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387187203196794e-05,
+      "loss": 1.0107,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386240709587208e-05,
+      "loss": 1.0045,
+      "step": 1641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385293508748994e-05,
+      "loss": 0.9752,
+      "step": 1642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938434560075351e-05,
+      "loss": 0.9617,
+      "step": 1643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383396985672187e-05,
+      "loss": 0.9897,
+      "step": 1644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382447663576495e-05,
+      "loss": 1.0124,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381497634537964e-05,
+      "loss": 0.9569,
+      "step": 1646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380546898628178e-05,
+      "loss": 0.9524,
+      "step": 1647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379595455918773e-05,
+      "loss": 1.0144,
+      "step": 1648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378643306481435e-05,
+      "loss": 0.5965,
+      "step": 1649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377690450387905e-05,
+      "loss": 0.9816,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376736887709982e-05,
+      "loss": 0.9403,
+      "step": 1651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375782618519514e-05,
+      "loss": 1.0192,
+      "step": 1652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93748276428884e-05,
+      "loss": 0.9818,
+      "step": 1653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373871960888594e-05,
+      "loss": 0.9586,
+      "step": 1654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937291557259211e-05,
+      "loss": 0.5671,
+      "step": 1655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371958478071e-05,
+      "loss": 0.9658,
+      "step": 1656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371000677397393e-05,
+      "loss": 0.9948,
+      "step": 1657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370042170643445e-05,
+      "loss": 1.0185,
+      "step": 1658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369082957881376e-05,
+      "loss": 0.9918,
+      "step": 1659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368123039183468e-05,
+      "loss": 0.9967,
+      "step": 1660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367162414622044e-05,
+      "loss": 0.9784,
+      "step": 1661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366201084269486e-05,
+      "loss": 1.0347,
+      "step": 1662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365239048198227e-05,
+      "loss": 1.0114,
+      "step": 1663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936427630648075e-05,
+      "loss": 0.9617,
+      "step": 1664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363312859189602e-05,
+      "loss": 0.985,
+      "step": 1665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362348706397374e-05,
+      "loss": 0.9876,
+      "step": 1666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936138384817671e-05,
+      "loss": 0.9896,
+      "step": 1667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360418284600313e-05,
+      "loss": 0.9744,
+      "step": 1668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935945201574093e-05,
+      "loss": 0.9618,
+      "step": 1669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935848504167137e-05,
+      "loss": 0.9737,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357517362464496e-05,
+      "loss": 1.0066,
+      "step": 1671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356548978193216e-05,
+      "loss": 0.965,
+      "step": 1672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355579888930494e-05,
+      "loss": 1.0041,
+      "step": 1673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354610094749353e-05,
+      "loss": 1.0444,
+      "step": 1674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353639595722863e-05,
+      "loss": 1.0014,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352668391924145e-05,
+      "loss": 0.9979,
+      "step": 1676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351696483426378e-05,
+      "loss": 0.9875,
+      "step": 1677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935072387030279e-05,
+      "loss": 0.9854,
+      "step": 1678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349750552626672e-05,
+      "loss": 0.9865,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934877653047136e-05,
+      "loss": 1.0028,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347801803910236e-05,
+      "loss": 1.0202,
+      "step": 1681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346826373016752e-05,
+      "loss": 1.01,
+      "step": 1682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93458502378644e-05,
+      "loss": 1.0399,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344873398526733e-05,
+      "loss": 1.0295,
+      "step": 1684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343895855077344e-05,
+      "loss": 0.9605,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342917607589897e-05,
+      "loss": 1.0326,
+      "step": 1686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341938656138097e-05,
+      "loss": 0.9917,
+      "step": 1687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340959000795707e-05,
+      "loss": 0.9964,
+      "step": 1688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339978641636542e-05,
+      "loss": 0.9205,
+      "step": 1689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338997578734466e-05,
+      "loss": 0.9773,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338015812163404e-05,
+      "loss": 0.9517,
+      "step": 1691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337033341997323e-05,
+      "loss": 0.972,
+      "step": 1692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933605016831026e-05,
+      "loss": 1.0493,
+      "step": 1693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335066291176284e-05,
+      "loss": 0.9784,
+      "step": 1694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334081710669533e-05,
+      "loss": 0.9571,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333096426864194e-05,
+      "loss": 0.9577,
+      "step": 1696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93321104398345e-05,
+      "loss": 1.0175,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331123749654748e-05,
+      "loss": 0.9695,
+      "step": 1698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933013635639928e-05,
+      "loss": 0.8991,
+      "step": 1699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932914826014249e-05,
+      "loss": 0.9836,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328159460958833e-05,
+      "loss": 0.9636,
+      "step": 1701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327169958922813e-05,
+      "loss": 0.9577,
+      "step": 1702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326179754108985e-05,
+      "loss": 0.9612,
+      "step": 1703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325188846591958e-05,
+      "loss": 1.0132,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324197236446397e-05,
+      "loss": 0.9974,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932320492374701e-05,
+      "loss": 0.9768,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322211908568572e-05,
+      "loss": 1.0167,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321218190985906e-05,
+      "loss": 0.9687,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320223771073877e-05,
+      "loss": 0.9831,
+      "step": 1709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319228648907414e-05,
+      "loss": 1.0179,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9318232824561507e-05,
+      "loss": 0.651,
+      "step": 1711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9317236298111176e-05,
+      "loss": 0.9533,
+      "step": 1712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9316239069631516e-05,
+      "loss": 0.9577,
+      "step": 1713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931524113919766e-05,
+      "loss": 0.9835,
+      "step": 1714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93142425068848e-05,
+      "loss": 1.0381,
+      "step": 1715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313243172768183e-05,
+      "loss": 0.9982,
+      "step": 1716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931224313692311e-05,
+      "loss": 0.9996,
+      "step": 1717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931124239942492e-05,
+      "loss": 0.9895,
+      "step": 1718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310240960349018e-05,
+      "loss": 0.9611,
+      "step": 1719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930923881977087e-05,
+      "loss": 0.9939,
+      "step": 1720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308235977765976e-05,
+      "loss": 1.0318,
+      "step": 1721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307232434409898e-05,
+      "loss": 0.9481,
+      "step": 1722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306228189778255e-05,
+      "loss": 1.0006,
+      "step": 1723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930522324394671e-05,
+      "loss": 0.9839,
+      "step": 1724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304217596990983e-05,
+      "loss": 1.0193,
+      "step": 1725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930321124898685e-05,
+      "loss": 0.9327,
+      "step": 1726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302204200010134e-05,
+      "loss": 0.5677,
+      "step": 1727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930119645013672e-05,
+      "loss": 0.9356,
+      "step": 1728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930018799944253e-05,
+      "loss": 0.9773,
+      "step": 1729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299178848003545e-05,
+      "loss": 0.6672,
+      "step": 1730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298168995895817e-05,
+      "loss": 1.011,
+      "step": 1731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929715844319542e-05,
+      "loss": 0.9706,
+      "step": 1732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929614718997851e-05,
+      "loss": 1.0335,
+      "step": 1733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295135236321278e-05,
+      "loss": 1.0138,
+      "step": 1734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294122582299964e-05,
+      "loss": 1.0153,
+      "step": 1735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929310922799088e-05,
+      "loss": 1.0099,
+      "step": 1736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292095173470373e-05,
+      "loss": 0.9973,
+      "step": 1737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291080418814852e-05,
+      "loss": 0.9631,
+      "step": 1738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290064964100773e-05,
+      "loss": 1.0174,
+      "step": 1739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289048809404653e-05,
+      "loss": 1.0171,
+      "step": 1740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928803195480305e-05,
+      "loss": 0.9455,
+      "step": 1741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287014400372584e-05,
+      "loss": 0.6185,
+      "step": 1742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285996146189932e-05,
+      "loss": 1.01,
+      "step": 1743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284977192331807e-05,
+      "loss": 0.9531,
+      "step": 1744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283957538874988e-05,
+      "loss": 1.015,
+      "step": 1745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92829371858963e-05,
+      "loss": 0.9611,
+      "step": 1746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281916133472636e-05,
+      "loss": 0.9649,
+      "step": 1747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280894381680917e-05,
+      "loss": 0.9983,
+      "step": 1748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927987193059813e-05,
+      "loss": 0.9721,
+      "step": 1749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927884878030132e-05,
+      "loss": 1.0232,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927782493086758e-05,
+      "loss": 0.9845,
+      "step": 1751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276800382374042e-05,
+      "loss": 1.0065,
+      "step": 1752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927577513489792e-05,
+      "loss": 1.0084,
+      "step": 1753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927474918851645e-05,
+      "loss": 1.009,
+      "step": 1754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927372254330694e-05,
+      "loss": 0.9948,
+      "step": 1755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272695199346743e-05,
+      "loss": 0.973,
+      "step": 1756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271667156713272e-05,
+      "loss": 0.9505,
+      "step": 1757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927063841548398e-05,
+      "loss": 0.9715,
+      "step": 1758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926960897573639e-05,
+      "loss": 0.9763,
+      "step": 1759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268578837548056e-05,
+      "loss": 0.9998,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92675480009966e-05,
+      "loss": 0.6651,
+      "step": 1761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9266516466159697e-05,
+      "loss": 0.9865,
+      "step": 1762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265484233115065e-05,
+      "loss": 0.9801,
+      "step": 1763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264451301940487e-05,
+      "loss": 0.9908,
+      "step": 1764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263417672713786e-05,
+      "loss": 0.9983,
+      "step": 1765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926238334551284e-05,
+      "loss": 1.0249,
+      "step": 1766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261348320415592e-05,
+      "loss": 0.9733,
+      "step": 1767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926031259750002e-05,
+      "loss": 1.0356,
+      "step": 1768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925927617684417e-05,
+      "loss": 1.0036,
+      "step": 1769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925823905852613e-05,
+      "loss": 0.9724,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257201242624045e-05,
+      "loss": 1.015,
+      "step": 1771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925616272921611e-05,
+      "loss": 1.0382,
+      "step": 1772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255123518380573e-05,
+      "loss": 0.9282,
+      "step": 1773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254083610195745e-05,
+      "loss": 1.0497,
+      "step": 1774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253043004739967e-05,
+      "loss": 0.9744,
+      "step": 1775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925200170209166e-05,
+      "loss": 0.9803,
+      "step": 1776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250959702329268e-05,
+      "loss": 1.024,
+      "step": 1777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249917005531315e-05,
+      "loss": 1.0095,
+      "step": 1778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924887361177636e-05,
+      "loss": 0.9922,
+      "step": 1779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247829521143023e-05,
+      "loss": 1.0159,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246784733709974e-05,
+      "loss": 0.5522,
+      "step": 1781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245739249555928e-05,
+      "loss": 1.0142,
+      "step": 1782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244693068759668e-05,
+      "loss": 0.9803,
+      "step": 1783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243646191400012e-05,
+      "loss": 0.5468,
+      "step": 1784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924259861755585e-05,
+      "loss": 1.0068,
+      "step": 1785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924155034730611e-05,
+      "loss": 0.9429,
+      "step": 1786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924050138072977e-05,
+      "loss": 0.986,
+      "step": 1787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239451717905878e-05,
+      "loss": 0.9568,
+      "step": 1788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238401358913513e-05,
+      "loss": 0.9733,
+      "step": 1789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237350303831822e-05,
+      "loss": 1.0091,
+      "step": 1790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236298552739998e-05,
+      "loss": 0.5476,
+      "step": 1791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923524610571729e-05,
+      "loss": 1.0199,
+      "step": 1792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234192962842996e-05,
+      "loss": 1.0132,
+      "step": 1793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233139124196464e-05,
+      "loss": 0.4961,
+      "step": 1794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232084589857103e-05,
+      "loss": 0.9502,
+      "step": 1795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231029359904366e-05,
+      "loss": 0.9662,
+      "step": 1796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922997343441776e-05,
+      "loss": 0.9929,
+      "step": 1797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228916813476855e-05,
+      "loss": 0.9622,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227859497161256e-05,
+      "loss": 0.9816,
+      "step": 1799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226801485550632e-05,
+      "loss": 1.0034,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92257427787247e-05,
+      "loss": 1.0355,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224683376763236e-05,
+      "loss": 0.9883,
+      "step": 1802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223623279746054e-05,
+      "loss": 0.9699,
+      "step": 1803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922256248775304e-05,
+      "loss": 1.0005,
+      "step": 1804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221501000864114e-05,
+      "loss": 1.024,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922043881915926e-05,
+      "loss": 0.9744,
+      "step": 1806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219375942718508e-05,
+      "loss": 0.9526,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218312371621944e-05,
+      "loss": 0.9974,
+      "step": 1808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217248105949708e-05,
+      "loss": 1.0311,
+      "step": 1809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216183145781984e-05,
+      "loss": 0.9798,
+      "step": 1810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921511749119902e-05,
+      "loss": 0.9931,
+      "step": 1811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214051142281107e-05,
+      "loss": 0.9565,
+      "step": 1812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212984099108594e-05,
+      "loss": 0.986,
+      "step": 1813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921191636176188e-05,
+      "loss": 1.0092,
+      "step": 1814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9210847930321407e-05,
+      "loss": 1.0107,
+      "step": 1815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920977880486769e-05,
+      "loss": 0.9697,
+      "step": 1816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920870898548128e-05,
+      "loss": 0.9162,
+      "step": 1817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207638472242785e-05,
+      "loss": 0.9846,
+      "step": 1818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206567265232867e-05,
+      "loss": 0.9916,
+      "step": 1819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920549536453224e-05,
+      "loss": 0.9786,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204422770221664e-05,
+      "loss": 0.9637,
+      "step": 1821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920334948238196e-05,
+      "loss": 0.9978,
+      "step": 1822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202275501093994e-05,
+      "loss": 0.9763,
+      "step": 1823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920120082643869e-05,
+      "loss": 1.0393,
+      "step": 1824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200125458497025e-05,
+      "loss": 0.9901,
+      "step": 1825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919904939735002e-05,
+      "loss": 1.0171,
+      "step": 1826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919797264307876e-05,
+      "loss": 0.995,
+      "step": 1827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196895195764363e-05,
+      "loss": 0.9843,
+      "step": 1828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919581705548802e-05,
+      "loss": 1.0011,
+      "step": 1829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919473822233097e-05,
+      "loss": 0.9631,
+      "step": 1830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9193658696374498e-05,
+      "loss": 0.9702,
+      "step": 1831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9192578477699938e-05,
+      "loss": 1.0015,
+      "step": 1832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191497566388685e-05,
+      "loss": 1.0238,
+      "step": 1833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190415962522186e-05,
+      "loss": 0.991,
+      "step": 1834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9189333666181933e-05,
+      "loss": 0.964,
+      "step": 1835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188250677449477e-05,
+      "loss": 0.9717,
+      "step": 1836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187166996406413e-05,
+      "loss": 1.0531,
+      "step": 1837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186082623134397e-05,
+      "loss": 1.0247,
+      "step": 1838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184997557715138e-05,
+      "loss": 0.9982,
+      "step": 1839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183911800230384e-05,
+      "loss": 1.0251,
+      "step": 1840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918282535076195e-05,
+      "loss": 1.015,
+      "step": 1841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181738209391695e-05,
+      "loss": 0.9959,
+      "step": 1842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180650376201536e-05,
+      "loss": 0.9798,
+      "step": 1843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179561851273433e-05,
+      "loss": 1.0081,
+      "step": 1844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178472634689407e-05,
+      "loss": 0.9876,
+      "step": 1845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177382726531527e-05,
+      "loss": 0.9847,
+      "step": 1846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917629212688191e-05,
+      "loss": 0.9897,
+      "step": 1847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917520083582274e-05,
+      "loss": 0.982,
+      "step": 1848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174108853436234e-05,
+      "loss": 1.0032,
+      "step": 1849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173016179804672e-05,
+      "loss": 0.9872,
+      "step": 1850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171922815010388e-05,
+      "loss": 0.9421,
+      "step": 1851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917082875913576e-05,
+      "loss": 0.9499,
+      "step": 1852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916973401226322e-05,
+      "loss": 0.985,
+      "step": 1853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168638574475264e-05,
+      "loss": 1.0176,
+      "step": 1854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916754244585442e-05,
+      "loss": 0.969,
+      "step": 1855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166445626483284e-05,
+      "loss": 0.9982,
+      "step": 1856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165348116444495e-05,
+      "loss": 0.9849,
+      "step": 1857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164249915820753e-05,
+      "loss": 0.9583,
+      "step": 1858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163151024694797e-05,
+      "loss": 0.9879,
+      "step": 1859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162051443149434e-05,
+      "loss": 0.9945,
+      "step": 1860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160951171267508e-05,
+      "loss": 1.0364,
+      "step": 1861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159850209131922e-05,
+      "loss": 1.0316,
+      "step": 1862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158748556825637e-05,
+      "loss": 0.5847,
+      "step": 1863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157646214431653e-05,
+      "loss": 0.9733,
+      "step": 1864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9156543182033028e-05,
+      "loss": 0.9882,
+      "step": 1865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9155439459712878e-05,
+      "loss": 0.9897,
+      "step": 1866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154335047554364e-05,
+      "loss": 1.0258,
+      "step": 1867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153229945640695e-05,
+      "loss": 0.9937,
+      "step": 1868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915212415405515e-05,
+      "loss": 0.9741,
+      "step": 1869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151017672881032e-05,
+      "loss": 1.0007,
+      "step": 1870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914991050220172e-05,
+      "loss": 0.9728,
+      "step": 1871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148802642100638e-05,
+      "loss": 0.9961,
+      "step": 1872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147694092661254e-05,
+      "loss": 0.9323,
+      "step": 1873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91465848539671e-05,
+      "loss": 0.9767,
+      "step": 1874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145474926101752e-05,
+      "loss": 0.9201,
+      "step": 1875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144364309148842e-05,
+      "loss": 0.9352,
+      "step": 1876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143253003192052e-05,
+      "loss": 1.0117,
+      "step": 1877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914214100831511e-05,
+      "loss": 0.9728,
+      "step": 1878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141028324601808e-05,
+      "loss": 0.9761,
+      "step": 1879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913991495213598e-05,
+      "loss": 0.9717,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913880089100152e-05,
+      "loss": 0.941,
+      "step": 1881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137686141282368e-05,
+      "loss": 0.9633,
+      "step": 1882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136570703062517e-05,
+      "loss": 1.0182,
+      "step": 1883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.9636,
+      "step": 1884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913433776145695e-05,
+      "loss": 1.0168,
+      "step": 1885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913322025823948e-05,
+      "loss": 0.9714,
+      "step": 1886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132102066857803e-05,
+      "loss": 0.9287,
+      "step": 1887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130983187396174e-05,
+      "loss": 1.0208,
+      "step": 1888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129863619938894e-05,
+      "loss": 0.9511,
+      "step": 1889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128743364570322e-05,
+      "loss": 0.9389,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127622421374866e-05,
+      "loss": 0.9312,
+      "step": 1891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126500790436985e-05,
+      "loss": 0.9723,
+      "step": 1892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125378471841195e-05,
+      "loss": 0.927,
+      "step": 1893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124255465672053e-05,
+      "loss": 0.529,
+      "step": 1894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123131772014178e-05,
+      "loss": 0.9962,
+      "step": 1895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122007390952238e-05,
+      "loss": 0.9472,
+      "step": 1896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120882322570952e-05,
+      "loss": 0.9762,
+      "step": 1897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119756566955092e-05,
+      "loss": 0.9923,
+      "step": 1898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118630124189476e-05,
+      "loss": 0.9684,
+      "step": 1899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117502994358984e-05,
+      "loss": 0.9969,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911637517754854e-05,
+      "loss": 0.5272,
+      "step": 1901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911524667384312e-05,
+      "loss": 0.9787,
+      "step": 1902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911411748332776e-05,
+      "loss": 0.9389,
+      "step": 1903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112987606087533e-05,
+      "loss": 1.0036,
+      "step": 1904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111857042207582e-05,
+      "loss": 1.0281,
+      "step": 1905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110725791773085e-05,
+      "loss": 1.0097,
+      "step": 1906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910959385486928e-05,
+      "loss": 1.0184,
+      "step": 1907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108461231581456e-05,
+      "loss": 0.9691,
+      "step": 1908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910732792199495e-05,
+      "loss": 1.025,
+      "step": 1909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910619392619516e-05,
+      "loss": 1.0012,
+      "step": 1910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105059244267526e-05,
+      "loss": 0.9593,
+      "step": 1911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103923876297544e-05,
+      "loss": 0.9341,
+      "step": 1912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102787822370764e-05,
+      "loss": 0.9637,
+      "step": 1913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910165108257278e-05,
+      "loss": 0.9691,
+      "step": 1914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100513656989244e-05,
+      "loss": 0.9971,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099375545705856e-05,
+      "loss": 0.9488,
+      "step": 1916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098236748808375e-05,
+      "loss": 0.9708,
+      "step": 1917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097097266382598e-05,
+      "loss": 1.0015,
+      "step": 1918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909595709851439e-05,
+      "loss": 1.0075,
+      "step": 1919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909481624528966e-05,
+      "loss": 0.9762,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093674706794363e-05,
+      "loss": 1.0169,
+      "step": 1921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9092532483114513e-05,
+      "loss": 1.0292,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091389574336173e-05,
+      "loss": 0.982,
+      "step": 1923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909024598054546e-05,
+      "loss": 0.9764,
+      "step": 1924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089101701828538e-05,
+      "loss": 0.9357,
+      "step": 1925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908795673827163e-05,
+      "loss": 0.9763,
+      "step": 1926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086811089961e-05,
+      "loss": 1.0035,
+      "step": 1927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085664756982975e-05,
+      "loss": 0.9617,
+      "step": 1928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084517739423923e-05,
+      "loss": 0.9672,
+      "step": 1929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083370037370276e-05,
+      "loss": 0.5712,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082221650908505e-05,
+      "loss": 0.9676,
+      "step": 1931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081072580125137e-05,
+      "loss": 1.0131,
+      "step": 1932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907992282510675e-05,
+      "loss": 1.0009,
+      "step": 1933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078772385939982e-05,
+      "loss": 0.9586,
+      "step": 1934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077621262711513e-05,
+      "loss": 0.9857,
+      "step": 1935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076469455508072e-05,
+      "loss": 1.0237,
+      "step": 1936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075316964416454e-05,
+      "loss": 0.9805,
+      "step": 1937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907416378952349e-05,
+      "loss": 0.9852,
+      "step": 1938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073009930916063e-05,
+      "loss": 0.954,
+      "step": 1939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071855388681125e-05,
+      "loss": 0.9928,
+      "step": 1940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070700162905663e-05,
+      "loss": 0.9628,
+      "step": 1941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906954425367672e-05,
+      "loss": 1.0069,
+      "step": 1942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068387661081388e-05,
+      "loss": 1.01,
+      "step": 1943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067230385206818e-05,
+      "loss": 0.9978,
+      "step": 1944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066072426140203e-05,
+      "loss": 0.9631,
+      "step": 1945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064913783968796e-05,
+      "loss": 0.9318,
+      "step": 1946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9063754458779898e-05,
+      "loss": 1.0081,
+      "step": 1947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9062594450660857e-05,
+      "loss": 1.0004,
+      "step": 1948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.906143375969908e-05,
+      "loss": 0.9923,
+      "step": 1949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9060272385982022e-05,
+      "loss": 0.9951,
+      "step": 1950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059110329597185e-05,
+      "loss": 0.9912,
+      "step": 1951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905794759063213e-05,
+      "loss": 1.0125,
+      "step": 1952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056784169174472e-05,
+      "loss": 0.915,
+      "step": 1953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905562006531186e-05,
+      "loss": 0.9461,
+      "step": 1954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905445527913202e-05,
+      "loss": 0.9966,
+      "step": 1955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90532898107227e-05,
+      "loss": 1.0018,
+      "step": 1956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905212366017173e-05,
+      "loss": 0.9572,
+      "step": 1957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905095682756697e-05,
+      "loss": 0.9958,
+      "step": 1958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904978931299633e-05,
+      "loss": 0.9636,
+      "step": 1959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9048621116547793e-05,
+      "loss": 0.9375,
+      "step": 1960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904745223830937e-05,
+      "loss": 1.0133,
+      "step": 1961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046282678369136e-05,
+      "loss": 1.0247,
+      "step": 1962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045112436815217e-05,
+      "loss": 0.9838,
+      "step": 1963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904394151373578e-05,
+      "loss": 0.9858,
+      "step": 1964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904276990921906e-05,
+      "loss": 0.9467,
+      "step": 1965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041597623353327e-05,
+      "loss": 1.0193,
+      "step": 1966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040424656226918e-05,
+      "loss": 0.9959,
+      "step": 1967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90392510079282e-05,
+      "loss": 0.9997,
+      "step": 1968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903807667854562e-05,
+      "loss": 0.9879,
+      "step": 1969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903690166816765e-05,
+      "loss": 0.9656,
+      "step": 1970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035725976882827e-05,
+      "loss": 1.0375,
+      "step": 1971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903454960477973e-05,
+      "loss": 1.0043,
+      "step": 1972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903337255194701e-05,
+      "loss": 1.0097,
+      "step": 1973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903219481847334e-05,
+      "loss": 0.9625,
+      "step": 1974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903101640444747e-05,
+      "loss": 1.0196,
+      "step": 1975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029837309958185e-05,
+      "loss": 1.0176,
+      "step": 1976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028657535094327e-05,
+      "loss": 0.9442,
+      "step": 1977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902747707994479e-05,
+      "loss": 0.9941,
+      "step": 1978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902629594459852e-05,
+      "loss": 1.0298,
+      "step": 1979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025114129144505e-05,
+      "loss": 0.948,
+      "step": 1980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90239316336718e-05,
+      "loss": 0.9432,
+      "step": 1981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90227484582695e-05,
+      "loss": 0.9927,
+      "step": 1982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021564603026756e-05,
+      "loss": 0.9907,
+      "step": 1983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902038006803276e-05,
+      "loss": 0.9722,
+      "step": 1984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019194853376776e-05,
+      "loss": 1.0218,
+      "step": 1985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018008959148097e-05,
+      "loss": 1.0015,
+      "step": 1986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016822385436086e-05,
+      "loss": 0.5429,
+      "step": 1987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901563513233014e-05,
+      "loss": 0.9503,
+      "step": 1988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014447199919715e-05,
+      "loss": 0.952,
+      "step": 1989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013258588294324e-05,
+      "loss": 0.9778,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012069297543527e-05,
+      "loss": 0.9645,
+      "step": 1991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901087932775693e-05,
+      "loss": 0.9848,
+      "step": 1992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.9814,
+      "step": 1993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900849735143503e-05,
+      "loss": 1.0094,
+      "step": 1994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007305345079207e-05,
+      "loss": 0.962,
+      "step": 1995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006112660046535e-05,
+      "loss": 1.0298,
+      "step": 1996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900491929642688e-05,
+      "loss": 1.0276,
+      "step": 1997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900372525431016e-05,
+      "loss": 1.0015,
+      "step": 1998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900253053378634e-05,
+      "loss": 0.9916,
+      "step": 1999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001335134945447e-05,
+      "loss": 0.9685,
+      "step": 2000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000139057877542e-05,
+      "loss": 0.992,
+      "step": 2001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998942302672753e-05,
+      "loss": 0.5787,
+      "step": 2002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997744869421248e-05,
+      "loss": 0.9908,
+      "step": 2003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996546758213248e-05,
+      "loss": 0.9529,
+      "step": 2004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995347969139034e-05,
+      "loss": 0.9816,
+      "step": 2005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899414850228893e-05,
+      "loss": 0.9846,
+      "step": 2006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992948357753314e-05,
+      "loss": 1.0007,
+      "step": 2007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991747535622607e-05,
+      "loss": 0.9818,
+      "step": 2008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990546035987296e-05,
+      "loss": 0.9886,
+      "step": 2009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989343858937906e-05,
+      "loss": 0.9718,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988141004565017e-05,
+      "loss": 0.9695,
+      "step": 2011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986937472959265e-05,
+      "loss": 0.937,
+      "step": 2012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985733264211332e-05,
+      "loss": 0.9521,
+      "step": 2013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898452837841195e-05,
+      "loss": 0.9638,
+      "step": 2014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983322815651903e-05,
+      "loss": 0.9566,
+      "step": 2015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982116576022028e-05,
+      "loss": 0.9879,
+      "step": 2016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980909659613217e-05,
+      "loss": 0.5886,
+      "step": 2017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979702066516402e-05,
+      "loss": 0.9434,
+      "step": 2018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897849379682257e-05,
+      "loss": 0.9986,
+      "step": 2019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897728485062277e-05,
+      "loss": 0.9742,
+      "step": 2020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976075228008087e-05,
+      "loss": 0.5556,
+      "step": 2021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974864929069662e-05,
+      "loss": 0.9468,
+      "step": 2022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897365395389869e-05,
+      "loss": 1.0074,
+      "step": 2023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972442302586413e-05,
+      "loss": 0.9384,
+      "step": 2024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8971229975224128e-05,
+      "loss": 0.9681,
+      "step": 2025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897001697190318e-05,
+      "loss": 0.9872,
+      "step": 2026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968803292714964e-05,
+      "loss": 0.9683,
+      "step": 2027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967588937750933e-05,
+      "loss": 1.0032,
+      "step": 2028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8966373907102577e-05,
+      "loss": 0.995,
+      "step": 2029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896515820086145e-05,
+      "loss": 0.9956,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963941819119155e-05,
+      "loss": 1.0022,
+      "step": 2031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896272476196734e-05,
+      "loss": 1.0098,
+      "step": 2032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961507029497703e-05,
+      "loss": 0.9918,
+      "step": 2033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960288621802003e-05,
+      "loss": 0.964,
+      "step": 2034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959069538972043e-05,
+      "loss": 0.9723,
+      "step": 2035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957849781099676e-05,
+      "loss": 0.9447,
+      "step": 2036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956629348276807e-05,
+      "loss": 1.0209,
+      "step": 2037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955408240595396e-05,
+      "loss": 0.9311,
+      "step": 2038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954186458147447e-05,
+      "loss": 0.9803,
+      "step": 2039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895296400102502e-05,
+      "loss": 1.0143,
+      "step": 2040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895174086932022e-05,
+      "loss": 0.9594,
+      "step": 2041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950517063125212e-05,
+      "loss": 0.9426,
+      "step": 2042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89492925825322e-05,
+      "loss": 0.981,
+      "step": 2043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948067427633456e-05,
+      "loss": 1.0387,
+      "step": 2044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946841598521285e-05,
+      "loss": 0.9563,
+      "step": 2045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945615095288047e-05,
+      "loss": 1.0133,
+      "step": 2046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944387918026162e-05,
+      "loss": 0.9793,
+      "step": 2047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943160066828093e-05,
+      "loss": 0.9621,
+      "step": 2048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941931541786353e-05,
+      "loss": 0.9455,
+      "step": 2049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940702342993512e-05,
+      "loss": 0.9421,
+      "step": 2050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939472470542185e-05,
+      "loss": 1.0008,
+      "step": 2051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938241924525037e-05,
+      "loss": 1.0481,
+      "step": 2052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937010705034788e-05,
+      "loss": 0.9204,
+      "step": 2053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893577881216421e-05,
+      "loss": 1.0193,
+      "step": 2054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934546246006122e-05,
+      "loss": 0.9928,
+      "step": 2055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933313006653392e-05,
+      "loss": 0.9472,
+      "step": 2056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932079094198944e-05,
+      "loss": 0.9666,
+      "step": 2057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930844508735746e-05,
+      "loss": 0.9295,
+      "step": 2058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929609250356827e-05,
+      "loss": 0.9597,
+      "step": 2059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928373319155255e-05,
+      "loss": 1.0474,
+      "step": 2060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892713671522416e-05,
+      "loss": 0.9694,
+      "step": 2061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925899438656708e-05,
+      "loss": 0.9898,
+      "step": 2062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924661489546136e-05,
+      "loss": 0.9844,
+      "step": 2063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892342286798571e-05,
+      "loss": 0.9302,
+      "step": 2064
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.892218357406876e-05,
+      "loss": 1.0149,
+      "step": 2065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8920943607888666e-05,
+      "loss": 0.9745,
+      "step": 2066
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8919702969538855e-05,
+      "loss": 0.9713,
+      "step": 2067
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8918461659112805e-05,
+      "loss": 1.0129,
+      "step": 2068
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8917219676704045e-05,
+      "loss": 0.9303,
+      "step": 2069
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915977022406156e-05,
+      "loss": 1.0083,
+      "step": 2070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891473369631277e-05,
+      "loss": 0.9354,
+      "step": 2071
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913489698517566e-05,
+      "loss": 1.0004,
+      "step": 2072
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891224502911428e-05,
+      "loss": 0.9969,
+      "step": 2073
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8910999688196688e-05,
+      "loss": 0.964,
+      "step": 2074
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890975367585863e-05,
+      "loss": 0.9972,
+      "step": 2075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908506992193984e-05,
+      "loss": 0.9462,
+      "step": 2076
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890725963729669e-05,
+      "loss": 0.9777,
+      "step": 2077
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890601161126073e-05,
+      "loss": 1.0304,
+      "step": 2078
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904762914180137e-05,
+      "loss": 0.9959,
+      "step": 2079
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903513546149e-05,
+      "loss": 0.9947,
+      "step": 2080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902263507261455e-05,
+      "loss": 0.9288,
+      "step": 2081
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8901012797611692e-05,
+      "loss": 1.0192,
+      "step": 2082
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899761417293944e-05,
+      "loss": 1.0005,
+      "step": 2083
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88985093664025e-05,
+      "loss": 0.9927,
+      "step": 2084
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88972566450317e-05,
+      "loss": 1.0217,
+      "step": 2085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896003253275934e-05,
+      "loss": 1.0038,
+      "step": 2086
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889474919122964e-05,
+      "loss": 0.9688,
+      "step": 2087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889349445898731e-05,
+      "loss": 0.9752,
+      "step": 2088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892239056643487e-05,
+      "loss": 0.9378,
+      "step": 2089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890982984292756e-05,
+      "loss": 1.0375,
+      "step": 2090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889726242029762e-05,
+      "loss": 0.982,
+      "step": 2091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88884688299492e-05,
+      "loss": 0.9753,
+      "step": 2092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887210748145806e-05,
+      "loss": 0.967,
+      "step": 2093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885951996714383e-05,
+      "loss": 0.9838,
+      "step": 2094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884692575749762e-05,
+      "loss": 0.9682,
+      "step": 2095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888343248534685e-05,
+      "loss": 1.002,
+      "step": 2096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888217172560058e-05,
+      "loss": 1.0362,
+      "step": 2097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880910296605956e-05,
+      "loss": 0.5997,
+      "step": 2098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887964819845802e-05,
+      "loss": 0.977,
+      "step": 2099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878385431251864e-05,
+      "loss": 0.9706,
+      "step": 2100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877121995082638e-05,
+      "loss": 1.0092,
+      "step": 2101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875857890045544e-05,
+      "loss": 1.0055,
+      "step": 2102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887459311623582e-05,
+      "loss": 0.9739,
+      "step": 2103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873327673748765e-05,
+      "loss": 0.9867,
+      "step": 2104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887206156267973e-05,
+      "loss": 1.0145,
+      "step": 2105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887079478312411e-05,
+      "loss": 0.9598,
+      "step": 2106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869527335177354e-05,
+      "loss": 0.9943,
+      "step": 2107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886825921893497e-05,
+      "loss": 0.9333,
+      "step": 2108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886699043449249e-05,
+      "loss": 1.0217,
+      "step": 2109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886572098194553e-05,
+      "loss": 0.9298,
+      "step": 2110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864450861389727e-05,
+      "loss": 0.9401,
+      "step": 2111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863180072920786e-05,
+      "loss": 0.5433,
+      "step": 2112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8861908616634465e-05,
+      "loss": 0.5354,
+      "step": 2113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860636492626555e-05,
+      "loss": 0.9834,
+      "step": 2114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885936370099291e-05,
+      "loss": 0.9798,
+      "step": 2115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858090241829435e-05,
+      "loss": 1.0064,
+      "step": 2116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856816115232076e-05,
+      "loss": 0.9765,
+      "step": 2117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885554132129684e-05,
+      "loss": 1.0018,
+      "step": 2118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854265860119777e-05,
+      "loss": 0.9922,
+      "step": 2119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885298973179699e-05,
+      "loss": 0.9846,
+      "step": 2120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885171293642463e-05,
+      "loss": 0.9725,
+      "step": 2121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850435474098903e-05,
+      "loss": 0.9367,
+      "step": 2122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849157344916066e-05,
+      "loss": 1.0172,
+      "step": 2123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847878548972417e-05,
+      "loss": 0.9836,
+      "step": 2124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846599086364307e-05,
+      "loss": 0.9803,
+      "step": 2125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884531895718815e-05,
+      "loss": 0.9731,
+      "step": 2126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884403816154039e-05,
+      "loss": 1.008,
+      "step": 2127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884275669951754e-05,
+      "loss": 0.9126,
+      "step": 2128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884147457121615e-05,
+      "loss": 0.9953,
+      "step": 2129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840191776732823e-05,
+      "loss": 0.994,
+      "step": 2130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883890831616422e-05,
+      "loss": 0.9753,
+      "step": 2131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8837624189607048e-05,
+      "loss": 0.9555,
+      "step": 2132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836339397158053e-05,
+      "loss": 0.9587,
+      "step": 2133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883505393891405e-05,
+      "loss": 0.9987,
+      "step": 2134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883376781497189e-05,
+      "loss": 0.9918,
+      "step": 2135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883248102542848e-05,
+      "loss": 0.947,
+      "step": 2136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831193570380773e-05,
+      "loss": 0.9866,
+      "step": 2137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882990544992578e-05,
+      "loss": 0.9876,
+      "step": 2138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882861666416056e-05,
+      "loss": 1.0141,
+      "step": 2139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882732721318221e-05,
+      "loss": 1.0077,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826037097087896e-05,
+      "loss": 0.9764,
+      "step": 2141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882474631597482e-05,
+      "loss": 0.9993,
+      "step": 2142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823454869940243e-05,
+      "loss": 1.0064,
+      "step": 2143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822162759081468e-05,
+      "loss": 0.9833,
+      "step": 2144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820869983495854e-05,
+      "loss": 0.9287,
+      "step": 2145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881957654328081e-05,
+      "loss": 0.9873,
+      "step": 2146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818282438533788e-05,
+      "loss": 0.9802,
+      "step": 2147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88169876693523e-05,
+      "loss": 0.98,
+      "step": 2148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815692235833903e-05,
+      "loss": 0.9451,
+      "step": 2149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814396138076202e-05,
+      "loss": 0.9633,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8813099376176858e-05,
+      "loss": 0.9232,
+      "step": 2151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8811801950233576e-05,
+      "loss": 0.9107,
+      "step": 2152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810503860344115e-05,
+      "loss": 0.6533,
+      "step": 2153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880920510660628e-05,
+      "loss": 0.9785,
+      "step": 2154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807905689117932e-05,
+      "loss": 1.0471,
+      "step": 2155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806605607976974e-05,
+      "loss": 0.955,
+      "step": 2156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805304863281372e-05,
+      "loss": 1.0017,
+      "step": 2157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880400345512913e-05,
+      "loss": 1.006,
+      "step": 2158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88027013836183e-05,
+      "loss": 0.9565,
+      "step": 2159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801398648846998e-05,
+      "loss": 0.9883,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800095250913378e-05,
+      "loss": 1.0098,
+      "step": 2161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8798791189915645e-05,
+      "loss": 0.9914,
+      "step": 2162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879748646595206e-05,
+      "loss": 0.9953,
+      "step": 2163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879618107912093e-05,
+      "loss": 0.976,
+      "step": 2164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8794875029520613e-05,
+      "loss": 0.9784,
+      "step": 2165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793568317249514e-05,
+      "loss": 0.973,
+      "step": 2166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792260942406093e-05,
+      "loss": 0.9402,
+      "step": 2167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790952905088855e-05,
+      "loss": 1.0135,
+      "step": 2168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878964420539636e-05,
+      "loss": 0.9875,
+      "step": 2169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788334843427213e-05,
+      "loss": 0.9787,
+      "step": 2170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787024819280073e-05,
+      "loss": 0.9491,
+      "step": 2171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785714133053643e-05,
+      "loss": 0.9703,
+      "step": 2172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784402784846683e-05,
+      "loss": 0.9717,
+      "step": 2173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783090774757993e-05,
+      "loss": 1.012,
+      "step": 2174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781778102886443e-05,
+      "loss": 0.9604,
+      "step": 2175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878046476933093e-05,
+      "loss": 0.9805,
+      "step": 2176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8779150774190412e-05,
+      "loss": 1.0309,
+      "step": 2177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777836117563894e-05,
+      "loss": 1.017,
+      "step": 2178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8776520799550432e-05,
+      "loss": 0.9842,
+      "step": 2179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775204820249136e-05,
+      "loss": 0.9583,
+      "step": 2180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773888179759155e-05,
+      "loss": 0.9893,
+      "step": 2181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87725708781797e-05,
+      "loss": 0.9712,
+      "step": 2182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8771252915610022e-05,
+      "loss": 0.9684,
+      "step": 2183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876993429214943e-05,
+      "loss": 0.9878,
+      "step": 2184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8768615007897274e-05,
+      "loss": 0.9709,
+      "step": 2185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876729506295296e-05,
+      "loss": 0.9846,
+      "step": 2186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765974457415947e-05,
+      "loss": 0.9579,
+      "step": 2187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764653191385737e-05,
+      "loss": 0.6284,
+      "step": 2188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763331264961882e-05,
+      "loss": 0.9888,
+      "step": 2189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762008678243987e-05,
+      "loss": 0.9744,
+      "step": 2190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87606854313317e-05,
+      "loss": 0.9389,
+      "step": 2191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759361524324736e-05,
+      "loss": 0.9822,
+      "step": 2192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758036957322838e-05,
+      "loss": 1.0057,
+      "step": 2193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875671173042581e-05,
+      "loss": 0.9759,
+      "step": 2194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875538584373351e-05,
+      "loss": 0.9522,
+      "step": 2195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754059297345836e-05,
+      "loss": 1.0569,
+      "step": 2196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752732091362737e-05,
+      "loss": 0.9903,
+      "step": 2197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751404225884218e-05,
+      "loss": 0.9883,
+      "step": 2198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750075701010332e-05,
+      "loss": 0.9735,
+      "step": 2199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748746516841173e-05,
+      "loss": 0.5533,
+      "step": 2200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87474166734769e-05,
+      "loss": 0.963,
+      "step": 2201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874608617101771e-05,
+      "loss": 0.9534,
+      "step": 2202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874475500956385e-05,
+      "loss": 0.9858,
+      "step": 2203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874342318921562e-05,
+      "loss": 0.9993,
+      "step": 2204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742090710073375e-05,
+      "loss": 0.9903,
+      "step": 2205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874075757223751e-05,
+      "loss": 0.9675,
+      "step": 2206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739423775808466e-05,
+      "loss": 0.9956,
+      "step": 2207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738089320886756e-05,
+      "loss": 0.9464,
+      "step": 2208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873675420757292e-05,
+      "loss": 0.9357,
+      "step": 2209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735418435967546e-05,
+      "loss": 0.9796,
+      "step": 2210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87340820061713e-05,
+      "loss": 0.983,
+      "step": 2211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732744918284866e-05,
+      "loss": 1.0358,
+      "step": 2212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873140717240899e-05,
+      "loss": 0.9507,
+      "step": 2213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873006876864447e-05,
+      "loss": 0.9875,
+      "step": 2214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728729707092156e-05,
+      "loss": 0.9784,
+      "step": 2215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727389987852933e-05,
+      "loss": 1.0108,
+      "step": 2216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8726049611027753e-05,
+      "loss": 0.9977,
+      "step": 2217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724708576717607e-05,
+      "loss": 0.9872,
+      "step": 2218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872336688502354e-05,
+      "loss": 0.9669,
+      "step": 2219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722024536046644e-05,
+      "loss": 0.9522,
+      "step": 2220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720681529888057e-05,
+      "loss": 0.9802,
+      "step": 2221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871933786664898e-05,
+      "loss": 0.9734,
+      "step": 2222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871799354643065e-05,
+      "loss": 0.9803,
+      "step": 2223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716648569334355e-05,
+      "loss": 0.9799,
+      "step": 2224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715302935461438e-05,
+      "loss": 0.9641,
+      "step": 2225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713956644913295e-05,
+      "loss": 0.9473,
+      "step": 2226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871260969779135e-05,
+      "loss": 0.9983,
+      "step": 2227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871126209419711e-05,
+      "loss": 0.9774,
+      "step": 2228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87099138342321e-05,
+      "loss": 0.9321,
+      "step": 2229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8708564917997917e-05,
+      "loss": 1.0315,
+      "step": 2230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8707215345596192e-05,
+      "loss": 0.9648,
+      "step": 2231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705865117128615e-05,
+      "loss": 0.955,
+      "step": 2232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870451423269692e-05,
+      "loss": 0.9803,
+      "step": 2233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703162692402895e-05,
+      "loss": 0.6291,
+      "step": 2234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701810496348373e-05,
+      "loss": 1.0029,
+      "step": 2235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870045764463524e-05,
+      "loss": 0.9839,
+      "step": 2236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869910413736543e-05,
+      "loss": 1.0224,
+      "step": 2237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697749974640924e-05,
+      "loss": 0.9846,
+      "step": 2238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696395156563753e-05,
+      "loss": 0.9567,
+      "step": 2239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695039683236006e-05,
+      "loss": 0.9839,
+      "step": 2240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869368355475981e-05,
+      "loss": 0.9537,
+      "step": 2241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692326771237344e-05,
+      "loss": 0.9763,
+      "step": 2242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869096933277084e-05,
+      "loss": 0.9847,
+      "step": 2243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868961123946258e-05,
+      "loss": 0.9972,
+      "step": 2244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688252491414886e-05,
+      "loss": 0.9522,
+      "step": 2245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8686893088730143e-05,
+      "loss": 1.0026,
+      "step": 2246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685533031510777e-05,
+      "loss": 0.93,
+      "step": 2247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684172319859258e-05,
+      "loss": 0.9469,
+      "step": 2248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682810953878124e-05,
+      "loss": 0.9131,
+      "step": 2249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681448933669942e-05,
+      "loss": 0.9179,
+      "step": 2250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680086259337337e-05,
+      "loss": 1.0111,
+      "step": 2251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678722930982984e-05,
+      "loss": 0.9485,
+      "step": 2252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867735894870961e-05,
+      "loss": 0.9779,
+      "step": 2253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867599431261998e-05,
+      "loss": 0.9507,
+      "step": 2254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674629022816925e-05,
+      "loss": 0.9271,
+      "step": 2255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867326307940331e-05,
+      "loss": 0.9656,
+      "step": 2256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867189648248205e-05,
+      "loss": 0.9379,
+      "step": 2257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670529232156128e-05,
+      "loss": 0.986,
+      "step": 2258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669161328528553e-05,
+      "loss": 1.0042,
+      "step": 2259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8667792771702397e-05,
+      "loss": 0.987,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666423561780773e-05,
+      "loss": 0.9815,
+      "step": 2261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866505369886685e-05,
+      "loss": 0.9842,
+      "step": 2262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663683183063846e-05,
+      "loss": 0.9835,
+      "step": 2263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662312014475023e-05,
+      "loss": 1.0073,
+      "step": 2264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662312014475023e-05,
+      "loss": 0.9628,
+      "step": 2265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8660940193203694e-05,
+      "loss": 0.9985,
+      "step": 2266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659567719353223e-05,
+      "loss": 0.907,
+      "step": 2267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8658194593027025e-05,
+      "loss": 0.581,
+      "step": 2268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656820814328557e-05,
+      "loss": 1.0461,
+      "step": 2269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655446383361332e-05,
+      "loss": 0.9961,
+      "step": 2270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865407130022891e-05,
+      "loss": 1.0152,
+      "step": 2271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652695565034898e-05,
+      "loss": 0.9892,
+      "step": 2272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8651319177882957e-05,
+      "loss": 0.626,
+      "step": 2273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649942138876792e-05,
+      "loss": 0.9582,
+      "step": 2274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864856444812016e-05,
+      "loss": 0.9926,
+      "step": 2275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864718610571687e-05,
+      "loss": 0.97,
+      "step": 2276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645807111770767e-05,
+      "loss": 1.0199,
+      "step": 2277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644427466385764e-05,
+      "loss": 1.0562,
+      "step": 2278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864304716966581e-05,
+      "loss": 0.9902,
+      "step": 2279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641666221714904e-05,
+      "loss": 1.0068,
+      "step": 2280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640284622637102e-05,
+      "loss": 0.9419,
+      "step": 2281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638902372536502e-05,
+      "loss": 0.9913,
+      "step": 2282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637519471517252e-05,
+      "loss": 1.0212,
+      "step": 2283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863613591968355e-05,
+      "loss": 0.9791,
+      "step": 2284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634751717139644e-05,
+      "loss": 1.0073,
+      "step": 2285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863336686398983e-05,
+      "loss": 1.0432,
+      "step": 2286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631981360338453e-05,
+      "loss": 0.8868,
+      "step": 2287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630595206289904e-05,
+      "loss": 0.9925,
+      "step": 2288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862920840194863e-05,
+      "loss": 0.9978,
+      "step": 2289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862782094741912e-05,
+      "loss": 0.9592,
+      "step": 2290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862643284280592e-05,
+      "loss": 0.931,
+      "step": 2291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8625044088213615e-05,
+      "loss": 0.5324,
+      "step": 2292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623654683746845e-05,
+      "loss": 0.9968,
+      "step": 2293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86222646295103e-05,
+      "loss": 0.9863,
+      "step": 2294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620873925608717e-05,
+      "loss": 0.9751,
+      "step": 2295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619482572146877e-05,
+      "loss": 0.9114,
+      "step": 2296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861809056922962e-05,
+      "loss": 0.9724,
+      "step": 2297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861669791696183e-05,
+      "loss": 0.9986,
+      "step": 2298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861530461544844e-05,
+      "loss": 0.4722,
+      "step": 2299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613910664794424e-05,
+      "loss": 0.9063,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.861251606510482e-05,
+      "loss": 0.9897,
+      "step": 2301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8611120816484708e-05,
+      "loss": 0.9054,
+      "step": 2302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8609724919039213e-05,
+      "loss": 0.9848,
+      "step": 2303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860832837287351e-05,
+      "loss": 0.526,
+      "step": 2304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8606931178092832e-05,
+      "loss": 0.9606,
+      "step": 2305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605533334802448e-05,
+      "loss": 0.9998,
+      "step": 2306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604134843107685e-05,
+      "loss": 1.0206,
+      "step": 2307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602735703113914e-05,
+      "loss": 1.0334,
+      "step": 2308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8601335914926558e-05,
+      "loss": 0.9781,
+      "step": 2309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599935478651082e-05,
+      "loss": 0.9215,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859853439439302e-05,
+      "loss": 0.9835,
+      "step": 2311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859713266225792e-05,
+      "loss": 0.9392,
+      "step": 2312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859573028235141e-05,
+      "loss": 0.9852,
+      "step": 2313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859432725477916e-05,
+      "loss": 0.9967,
+      "step": 2314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592923579646874e-05,
+      "loss": 0.9413,
+      "step": 2315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859151925706032e-05,
+      "loss": 0.9491,
+      "step": 2316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590114287125313e-05,
+      "loss": 0.9465,
+      "step": 2317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858870866994771e-05,
+      "loss": 1.0,
+      "step": 2318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858730240563342e-05,
+      "loss": 0.9454,
+      "step": 2319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85858954942884e-05,
+      "loss": 0.9511,
+      "step": 2320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584487936018663e-05,
+      "loss": 0.9254,
+      "step": 2321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583079730930264e-05,
+      "loss": 0.9592,
+      "step": 2322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85816708791293e-05,
+      "loss": 0.9788,
+      "step": 2323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8580261380721932e-05,
+      "loss": 0.9493,
+      "step": 2324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857885123581436e-05,
+      "loss": 0.9957,
+      "step": 2325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857744044451283e-05,
+      "loss": 0.9645,
+      "step": 2326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576029006923653e-05,
+      "loss": 0.9536,
+      "step": 2327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574616923153163e-05,
+      "loss": 0.9734,
+      "step": 2328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573204193307767e-05,
+      "loss": 0.969,
+      "step": 2329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857179081749391e-05,
+      "loss": 0.9202,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570376795818078e-05,
+      "loss": 1.0142,
+      "step": 2331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568962128386817e-05,
+      "loss": 0.9961,
+      "step": 2332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567546815306726e-05,
+      "loss": 1.0044,
+      "step": 2333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566130856684437e-05,
+      "loss": 0.9987,
+      "step": 2334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856471425262664e-05,
+      "loss": 1.0163,
+      "step": 2335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563297003240078e-05,
+      "loss": 0.9391,
+      "step": 2336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561879108631533e-05,
+      "loss": 0.977,
+      "step": 2337
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560460568907836e-05,
+      "loss": 0.9607,
+      "step": 2338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8559041384175874e-05,
+      "loss": 0.9526,
+      "step": 2339
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855762155454258e-05,
+      "loss": 0.9664,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556201080114934e-05,
+      "loss": 0.9894,
+      "step": 2341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554779960999963e-05,
+      "loss": 0.9464,
+      "step": 2342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8553358197304744e-05,
+      "loss": 0.9933,
+      "step": 2343
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855193578913641e-05,
+      "loss": 0.9397,
+      "step": 2344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550512736602125e-05,
+      "loss": 1.0386,
+      "step": 2345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854908903980912e-05,
+      "loss": 1.0361,
+      "step": 2346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547664698864666e-05,
+      "loss": 0.9341,
+      "step": 2347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8546239713876077e-05,
+      "loss": 0.9682,
+      "step": 2348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854481408495073e-05,
+      "loss": 0.9709,
+      "step": 2349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854338781219604e-05,
+      "loss": 0.9553,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8541960895719473e-05,
+      "loss": 0.9667,
+      "step": 2351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540533335628542e-05,
+      "loss": 0.9729,
+      "step": 2352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853910513203081e-05,
+      "loss": 0.9705,
+      "step": 2353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537676285033886e-05,
+      "loss": 0.9443,
+      "step": 2354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536246794745437e-05,
+      "loss": 0.98,
+      "step": 2355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534816661273165e-05,
+      "loss": 0.9852,
+      "step": 2356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853338588472483e-05,
+      "loss": 0.9499,
+      "step": 2357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531954465208234e-05,
+      "loss": 0.9796,
+      "step": 2358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530522402831236e-05,
+      "loss": 0.9459,
+      "step": 2359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529089697701735e-05,
+      "loss": 0.959,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852765634992768e-05,
+      "loss": 0.9385,
+      "step": 2361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526222359617073e-05,
+      "loss": 0.9633,
+      "step": 2362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524787726877956e-05,
+      "loss": 0.9399,
+      "step": 2363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523352451818435e-05,
+      "loss": 0.8967,
+      "step": 2364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521916534546643e-05,
+      "loss": 0.9404,
+      "step": 2365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520479975170783e-05,
+      "loss": 0.9916,
+      "step": 2366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8519042773799087e-05,
+      "loss": 1.0204,
+      "step": 2367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851760493053985e-05,
+      "loss": 0.9842,
+      "step": 2368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516166445501405e-05,
+      "loss": 1.0035,
+      "step": 2369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851472731879214e-05,
+      "loss": 0.9668,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513287550520494e-05,
+      "loss": 0.9753,
+      "step": 2371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8511847140794944e-05,
+      "loss": 0.9987,
+      "step": 2372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510406089724026e-05,
+      "loss": 0.9974,
+      "step": 2373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508964397416312e-05,
+      "loss": 0.9824,
+      "step": 2374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507522063980436e-05,
+      "loss": 1.0275,
+      "step": 2375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506079089525072e-05,
+      "loss": 0.9483,
+      "step": 2376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850463547415895e-05,
+      "loss": 0.9949,
+      "step": 2377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850319121799083e-05,
+      "loss": 0.9651,
+      "step": 2378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501746321129542e-05,
+      "loss": 0.9696,
+      "step": 2379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500300783683956e-05,
+      "loss": 0.9327,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498854605762982e-05,
+      "loss": 0.9353,
+      "step": 2381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8497407787475597e-05,
+      "loss": 1.0081,
+      "step": 2382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495960328930806e-05,
+      "loss": 0.9318,
+      "step": 2383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849451223023767e-05,
+      "loss": 0.9761,
+      "step": 2384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493063491505307e-05,
+      "loss": 1.0015,
+      "step": 2385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849161411284287e-05,
+      "loss": 1.012,
+      "step": 2386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849016409435957e-05,
+      "loss": 0.9428,
+      "step": 2387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848871343616466e-05,
+      "loss": 0.9907,
+      "step": 2388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848726213836744e-05,
+      "loss": 0.5875,
+      "step": 2389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485810201077266e-05,
+      "loss": 0.9293,
+      "step": 2390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848435762440354e-05,
+      "loss": 0.9645,
+      "step": 2391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84829044084557e-05,
+      "loss": 0.9223,
+      "step": 2392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848145055334325e-05,
+      "loss": 1.0015,
+      "step": 2393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8479996059175733e-05,
+      "loss": 1.022,
+      "step": 2394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478540926062743e-05,
+      "loss": 0.9713,
+      "step": 2395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477085154113913e-05,
+      "loss": 0.9885,
+      "step": 2396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475628743438944e-05,
+      "loss": 0.9703,
+      "step": 2397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474171694147557e-05,
+      "loss": 0.9684,
+      "step": 2398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8472714006349554e-05,
+      "loss": 0.9801,
+      "step": 2399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471255680154752e-05,
+      "loss": 0.9047,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469796715673045e-05,
+      "loss": 0.9695,
+      "step": 2401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846833711301435e-05,
+      "loss": 0.9649,
+      "step": 2402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466876872288653e-05,
+      "loss": 0.9104,
+      "step": 2403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846541599360598e-05,
+      "loss": 1.0268,
+      "step": 2404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84639544770764e-05,
+      "loss": 0.9809,
+      "step": 2405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462492322810033e-05,
+      "loss": 0.9978,
+      "step": 2406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461029530917054e-05,
+      "loss": 0.9935,
+      "step": 2407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459566101507675e-05,
+      "loss": 0.9401,
+      "step": 2408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845810203469217e-05,
+      "loss": 0.9596,
+      "step": 2409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8456637330580842e-05,
+      "loss": 0.9934,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845517198928406e-05,
+      "loss": 0.9862,
+      "step": 2411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845370601091223e-05,
+      "loss": 0.9709,
+      "step": 2412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845223939557581e-05,
+      "loss": 0.9149,
+      "step": 2413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845077214338531e-05,
+      "loss": 0.965,
+      "step": 2414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8449304254451273e-05,
+      "loss": 0.9607,
+      "step": 2415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447835728884315e-05,
+      "loss": 0.9578,
+      "step": 2416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446366566795075e-05,
+      "loss": 0.9464,
+      "step": 2417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444896768294255e-05,
+      "loss": 0.9943,
+      "step": 2418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84434263334926e-05,
+      "loss": 0.9995,
+      "step": 2419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84419552625009e-05,
+      "loss": 0.9846,
+      "step": 2420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.844048355543e-05,
+      "loss": 0.98,
+      "step": 2421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843901121239079e-05,
+      "loss": 0.9319,
+      "step": 2422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8437538233494205e-05,
+      "loss": 0.961,
+      "step": 2423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8436064618851225e-05,
+      "loss": 0.9828,
+      "step": 2424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434590368572896e-05,
+      "loss": 0.9781,
+      "step": 2425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433115482770286e-05,
+      "loss": 0.9278,
+      "step": 2426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843163996155453e-05,
+      "loss": 0.9324,
+      "step": 2427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430163805036807e-05,
+      "loss": 0.9625,
+      "step": 2428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428687013328338e-05,
+      "loss": 0.9611,
+      "step": 2429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427209586540392e-05,
+      "loss": 0.9846,
+      "step": 2430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425731524784297e-05,
+      "loss": 0.9418,
+      "step": 2431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8424252828171415e-05,
+      "loss": 1.0009,
+      "step": 2432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8422773496813164e-05,
+      "loss": 0.9915,
+      "step": 2433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842129353082101e-05,
+      "loss": 0.9976,
+      "step": 2434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8419812930306456e-05,
+      "loss": 1.0194,
+      "step": 2435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418331695381074e-05,
+      "loss": 0.937,
+      "step": 2436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416849826156462e-05,
+      "loss": 0.943,
+      "step": 2437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841536732274428e-05,
+      "loss": 0.9891,
+      "step": 2438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413884185256225e-05,
+      "loss": 0.9457,
+      "step": 2439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412400413804055e-05,
+      "loss": 0.9294,
+      "step": 2440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410916008499563e-05,
+      "loss": 0.9258,
+      "step": 2441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409430969454594e-05,
+      "loss": 0.9711,
+      "step": 2442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407945296781046e-05,
+      "loss": 0.9846,
+      "step": 2443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840645899059086e-05,
+      "loss": 0.9942,
+      "step": 2444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404972050996023e-05,
+      "loss": 0.9856,
+      "step": 2445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840348447810857e-05,
+      "loss": 0.9483,
+      "step": 2446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401996272040593e-05,
+      "loss": 0.9523,
+      "step": 2447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840050743290422e-05,
+      "loss": 0.9811,
+      "step": 2448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399017960811628e-05,
+      "loss": 1.001,
+      "step": 2449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839752785587505e-05,
+      "loss": 1.0078,
+      "step": 2450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839603711820676e-05,
+      "loss": 0.9459,
+      "step": 2451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839454574791908e-05,
+      "loss": 0.9843,
+      "step": 2452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839305374512438e-05,
+      "loss": 0.9738,
+      "step": 2453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391561109935083e-05,
+      "loss": 1.0339,
+      "step": 2454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8390067842463653e-05,
+      "loss": 1.0088,
+      "step": 2455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388573942822597e-05,
+      "loss": 1.0008,
+      "step": 2456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387079411124488e-05,
+      "loss": 1.0001,
+      "step": 2457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8385584247481928e-05,
+      "loss": 0.9517,
+      "step": 2458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838408845200758e-05,
+      "loss": 0.9591,
+      "step": 2459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382592024814136e-05,
+      "loss": 1.0006,
+      "step": 2460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838109496601436e-05,
+      "loss": 0.937,
+      "step": 2461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379597275721043e-05,
+      "loss": 0.9587,
+      "step": 2462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837809895404704e-05,
+      "loss": 0.9714,
+      "step": 2463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837660000110524e-05,
+      "loss": 0.9001,
+      "step": 2464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8375100417008584e-05,
+      "loss": 0.9633,
+      "step": 2465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8373600201870062e-05,
+      "loss": 0.8947,
+      "step": 2466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837209935580272e-05,
+      "loss": 0.9911,
+      "step": 2467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370597878919633e-05,
+      "loss": 0.9917,
+      "step": 2468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369095771333938e-05,
+      "loss": 1.0335,
+      "step": 2469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367593033158813e-05,
+      "loss": 0.9918,
+      "step": 2470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8366089664507488e-05,
+      "loss": 0.9268,
+      "step": 2471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364585665493234e-05,
+      "loss": 0.9574,
+      "step": 2472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8363081036229373e-05,
+      "loss": 0.9661,
+      "step": 2473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836157577682928e-05,
+      "loss": 0.9448,
+      "step": 2474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8360069887406372e-05,
+      "loss": 0.9483,
+      "step": 2475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358563368074108e-05,
+      "loss": 0.9383,
+      "step": 2476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8357056218946003e-05,
+      "loss": 0.8992,
+      "step": 2477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835554844013562e-05,
+      "loss": 0.9583,
+      "step": 2478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835404003175656e-05,
+      "loss": 1.0075,
+      "step": 2479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352530993922483e-05,
+      "loss": 0.9536,
+      "step": 2480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835102132674709e-05,
+      "loss": 0.981,
+      "step": 2481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834951103034413e-05,
+      "loss": 0.9311,
+      "step": 2482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348000104827396e-05,
+      "loss": 0.9569,
+      "step": 2483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834648855031074e-05,
+      "loss": 0.9806,
+      "step": 2484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344976366908048e-05,
+      "loss": 1.011,
+      "step": 2485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834346355473326e-05,
+      "loss": 1.0025,
+      "step": 2486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341950113900363e-05,
+      "loss": 0.9743,
+      "step": 2487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340436044523392e-05,
+      "loss": 0.9559,
+      "step": 2488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338921346716426e-05,
+      "loss": 0.9258,
+      "step": 2489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337406020593595e-05,
+      "loss": 0.9445,
+      "step": 2490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833589006626907e-05,
+      "loss": 0.8903,
+      "step": 2491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334373483857083e-05,
+      "loss": 0.9464,
+      "step": 2492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83328562734719e-05,
+      "loss": 1.0171,
+      "step": 2493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331338435227838e-05,
+      "loss": 0.9802,
+      "step": 2494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329819969239263e-05,
+      "loss": 0.9529,
+      "step": 2495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328300875620586e-05,
+      "loss": 0.5296,
+      "step": 2496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8326781154486274e-05,
+      "loss": 0.968,
+      "step": 2497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832526080595082e-05,
+      "loss": 1.0002,
+      "step": 2498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8323739830128787e-05,
+      "loss": 0.95,
+      "step": 2499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832221822713478e-05,
+      "loss": 0.962,
+      "step": 2500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320695997083443e-05,
+      "loss": 0.9447,
+      "step": 2501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8319173140089473e-05,
+      "loss": 1.0048,
+      "step": 2502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317649656267612e-05,
+      "loss": 1.006,
+      "step": 2503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8316125545732653e-05,
+      "loss": 0.9842,
+      "step": 2504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831460080859943e-05,
+      "loss": 0.9659,
+      "step": 2505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313075444982834e-05,
+      "loss": 1.0313,
+      "step": 2506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311549454997788e-05,
+      "loss": 0.9506,
+      "step": 2507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831002283875928e-05,
+      "loss": 0.9565,
+      "step": 2508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830849559638233e-05,
+      "loss": 0.9322,
+      "step": 2509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830696772798202e-05,
+      "loss": 0.9786,
+      "step": 2510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305439233673462e-05,
+      "loss": 0.9464,
+      "step": 2511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830391011357183e-05,
+      "loss": 0.9546,
+      "step": 2512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8302380367792336e-05,
+      "loss": 0.9916,
+      "step": 2513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300849996450244e-05,
+      "loss": 0.9401,
+      "step": 2514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299318999660863e-05,
+      "loss": 0.9156,
+      "step": 2515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829778737753955e-05,
+      "loss": 0.9902,
+      "step": 2516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8296255130201707e-05,
+      "loss": 0.9359,
+      "step": 2517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294722257762788e-05,
+      "loss": 0.9686,
+      "step": 2518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293188760338285e-05,
+      "loss": 0.9533,
+      "step": 2519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291654638043746e-05,
+      "loss": 0.9988,
+      "step": 2520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290119890994772e-05,
+      "loss": 0.9812,
+      "step": 2521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8288584519306985e-05,
+      "loss": 0.9795,
+      "step": 2522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828704852309609e-05,
+      "loss": 0.9589,
+      "step": 2523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285511902477803e-05,
+      "loss": 0.9553,
+      "step": 2524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283974657567915e-05,
+      "loss": 0.9518,
+      "step": 2525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828243678848225e-05,
+      "loss": 0.9591,
+      "step": 2526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280898295336683e-05,
+      "loss": 0.9358,
+      "step": 2527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279359178247134e-05,
+      "loss": 0.9784,
+      "step": 2528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277819437329577e-05,
+      "loss": 0.9524,
+      "step": 2529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8276279072700017e-05,
+      "loss": 0.9785,
+      "step": 2530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8274738084474525e-05,
+      "loss": 0.9803,
+      "step": 2531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827319647276921e-05,
+      "loss": 0.9642,
+      "step": 2532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8271654237700224e-05,
+      "loss": 0.8991,
+      "step": 2533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270111379383773e-05,
+      "loss": 0.9414,
+      "step": 2534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826856789793611e-05,
+      "loss": 0.931,
+      "step": 2535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826702379347353e-05,
+      "loss": 0.6026,
+      "step": 2536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.826547906611237e-05,
+      "loss": 0.9372,
+      "step": 2537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8263933715969032e-05,
+      "loss": 0.954,
+      "step": 2538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 0.6475,
+      "step": 2539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.826084114780161e-05,
+      "loss": 0.9442,
+      "step": 2540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825929393001054e-05,
+      "loss": 0.9885,
+      "step": 2541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825774608990332e-05,
+      "loss": 1.0185,
+      "step": 2542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256197627596582e-05,
+      "loss": 0.9579,
+      "step": 2543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825464854320699e-05,
+      "loss": 0.9886,
+      "step": 2544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825309883685127e-05,
+      "loss": 0.9476,
+      "step": 2545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251548508646186e-05,
+      "loss": 1.0077,
+      "step": 2546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824999755870855e-05,
+      "loss": 0.981,
+      "step": 2547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248445987155222e-05,
+      "loss": 0.9996,
+      "step": 2548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8246893794103113e-05,
+      "loss": 0.9763,
+      "step": 2549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245340979669165e-05,
+      "loss": 0.9669,
+      "step": 2550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243787543970394e-05,
+      "loss": 0.9713,
+      "step": 2551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242233487123837e-05,
+      "loss": 0.973,
+      "step": 2552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824067880924659e-05,
+      "loss": 0.9871,
+      "step": 2553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82391235104558e-05,
+      "loss": 0.5964,
+      "step": 2554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237567590868645e-05,
+      "loss": 0.973,
+      "step": 2555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236011050602368e-05,
+      "loss": 0.94,
+      "step": 2556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234453889774248e-05,
+      "loss": 0.9829,
+      "step": 2557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232896108501606e-05,
+      "loss": 0.9342,
+      "step": 2558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231337706901823e-05,
+      "loss": 0.9813,
+      "step": 2559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8229778685092322e-05,
+      "loss": 0.9482,
+      "step": 2560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822821904319057e-05,
+      "loss": 0.94,
+      "step": 2561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226658781314078e-05,
+      "loss": 0.9573,
+      "step": 2562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822509789958041e-05,
+      "loss": 0.9484,
+      "step": 2563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223536398107177e-05,
+      "loss": 0.9639,
+      "step": 2564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822197427701203e-05,
+      "loss": 1.0591,
+      "step": 2565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220411536412673e-05,
+      "loss": 0.9688,
+      "step": 2566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8218848176426857e-05,
+      "loss": 0.9627,
+      "step": 2567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217284197172372e-05,
+      "loss": 0.9437,
+      "step": 2568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821571959876706e-05,
+      "loss": 0.9814,
+      "step": 2569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8214154381328815e-05,
+      "loss": 0.9825,
+      "step": 2570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212588544975567e-05,
+      "loss": 0.9572,
+      "step": 2571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211022089825295e-05,
+      "loss": 0.9828,
+      "step": 2572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209455015996034e-05,
+      "loss": 0.9843,
+      "step": 2573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207887323605858e-05,
+      "loss": 0.9582,
+      "step": 2574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206319012772885e-05,
+      "loss": 0.9853,
+      "step": 2575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204750083615283e-05,
+      "loss": 0.9741,
+      "step": 2576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820318053625127e-05,
+      "loss": 0.8931,
+      "step": 2577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8201610370799104e-05,
+      "loss": 0.9266,
+      "step": 2578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82000395873771e-05,
+      "loss": 0.9705,
+      "step": 2579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198468186103602e-05,
+      "loss": 1.0054,
+      "step": 2580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819689616709702e-05,
+      "loss": 0.9866,
+      "step": 2581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819532353047579e-05,
+      "loss": 0.9935,
+      "step": 2582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819375027635842e-05,
+      "loss": 0.9463,
+      "step": 2583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192176404863442e-05,
+      "loss": 0.9673,
+      "step": 2584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190601916109444e-05,
+      "loss": 0.9975,
+      "step": 2585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189026810215064e-05,
+      "loss": 0.9911,
+      "step": 2586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187451087298974e-05,
+      "loss": 0.9623,
+      "step": 2587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185874747479912e-05,
+      "loss": 0.9671,
+      "step": 2588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184297790876638e-05,
+      "loss": 0.6004,
+      "step": 2589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818272021760798e-05,
+      "loss": 0.9971,
+      "step": 2590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181142027792807e-05,
+      "loss": 0.9789,
+      "step": 2591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817956322155002e-05,
+      "loss": 1.0096,
+      "step": 2592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8177983798998588e-05,
+      "loss": 0.9675,
+      "step": 2593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817640376025751e-05,
+      "loss": 0.9697,
+      "step": 2594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174823105445845e-05,
+      "loss": 1.0291,
+      "step": 2595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817324183468268e-05,
+      "loss": 0.9587,
+      "step": 2596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171659948087175e-05,
+      "loss": 0.9806,
+      "step": 2597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170077445778503e-05,
+      "loss": 0.9839,
+      "step": 2598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168494327875918e-05,
+      "loss": 0.9149,
+      "step": 2599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8166910594498695e-05,
+      "loss": 0.9522,
+      "step": 2600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165326245766164e-05,
+      "loss": 1.007,
+      "step": 2601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163741281797706e-05,
+      "loss": 0.94,
+      "step": 2602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816215570271274e-05,
+      "loss": 0.9699,
+      "step": 2603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160569508630734e-05,
+      "loss": 0.9473,
+      "step": 2604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8158982699671208e-05,
+      "loss": 0.9672,
+      "step": 2605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157395275953722e-05,
+      "loss": 0.9555,
+      "step": 2606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8155807237597884e-05,
+      "loss": 1.0208,
+      "step": 2607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815421858472335e-05,
+      "loss": 0.9574,
+      "step": 2608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152629317449814e-05,
+      "loss": 0.9629,
+      "step": 2609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151039435897032e-05,
+      "loss": 0.9898,
+      "step": 2610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149448940184797e-05,
+      "loss": 0.9868,
+      "step": 2611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814785783043294e-05,
+      "loss": 0.9156,
+      "step": 2612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8146266106761357e-05,
+      "loss": 0.9188,
+      "step": 2613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8144673769289972e-05,
+      "loss": 0.6183,
+      "step": 2614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814308081813877e-05,
+      "loss": 0.9334,
+      "step": 2615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8141487253427775e-05,
+      "loss": 0.5564,
+      "step": 2616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813989307527705e-05,
+      "loss": 0.9817,
+      "step": 2617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813829828380672e-05,
+      "loss": 0.9291,
+      "step": 2618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813670287913695e-05,
+      "loss": 0.9396,
+      "step": 2619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8135106861387945e-05,
+      "loss": 0.9411,
+      "step": 2620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8133510230679956e-05,
+      "loss": 0.9978,
+      "step": 2621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131912987133296e-05,
+      "loss": 0.9777,
+      "step": 2622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8130315130868304e-05,
+      "loss": 0.5514,
+      "step": 2623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8128716662005385e-05,
+      "loss": 1.0107,
+      "step": 2624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8127117580664966e-05,
+      "loss": 1.0039,
+      "step": 2625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812551788696754e-05,
+      "loss": 0.9732,
+      "step": 2626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123917581033642e-05,
+      "loss": 1.009,
+      "step": 2627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8122316662983846e-05,
+      "loss": 0.9618,
+      "step": 2628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120715132938784e-05,
+      "loss": 0.5914,
+      "step": 2629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8119112991019123e-05,
+      "loss": 0.9575,
+      "step": 2630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117510237345572e-05,
+      "loss": 0.9251,
+      "step": 2631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115906872038912e-05,
+      "loss": 0.9384,
+      "step": 2632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114302895219936e-05,
+      "loss": 0.9746,
+      "step": 2633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112698307009506e-05,
+      "loss": 0.9141,
+      "step": 2634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111093107528526e-05,
+      "loss": 0.9433,
+      "step": 2635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8109487296897944e-05,
+      "loss": 0.9858,
+      "step": 2636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107880875238746e-05,
+      "loss": 0.9898,
+      "step": 2637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106273842671974e-05,
+      "loss": 0.9364,
+      "step": 2638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810466619931872e-05,
+      "loss": 1.0094,
+      "step": 2639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8103057945300115e-05,
+      "loss": 0.9521,
+      "step": 2640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101449080737327e-05,
+      "loss": 0.9893,
+      "step": 2641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809983960575159e-05,
+      "loss": 0.9593,
+      "step": 2642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8098229520464167e-05,
+      "loss": 0.9869,
+      "step": 2643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809661882499638e-05,
+      "loss": 0.9646,
+      "step": 2644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8095007519469583e-05,
+      "loss": 0.9691,
+      "step": 2645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093395604005197e-05,
+      "loss": 0.9474,
+      "step": 2646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8091783078724657e-05,
+      "loss": 1.0011,
+      "step": 2647
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.9895,
+      "step": 2648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8088556199201195e-05,
+      "loss": 0.9534,
+      "step": 2649
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086941845201408e-05,
+      "loss": 0.9381,
+      "step": 2650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085326881871747e-05,
+      "loss": 0.9368,
+      "step": 2651
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80837113093339e-05,
+      "loss": 0.966,
+      "step": 2652
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8082095127709594e-05,
+      "loss": 0.9158,
+      "step": 2653
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080478337120604e-05,
+      "loss": 0.9819,
+      "step": 2654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8078860937688757e-05,
+      "loss": 0.9537,
+      "step": 2655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8077242929535913e-05,
+      "loss": 0.9372,
+      "step": 2656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8075624312783984e-05,
+      "loss": 0.5823,
+      "step": 2657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807400508755493e-05,
+      "loss": 1.0066,
+      "step": 2658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8072385253970756e-05,
+      "loss": 0.9792,
+      "step": 2659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8070764812153518e-05,
+      "loss": 0.9647,
+      "step": 2660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80691437622253e-05,
+      "loss": 1.0187,
+      "step": 2661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806752210430825e-05,
+      "loss": 0.9523,
+      "step": 2662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806589983852456e-05,
+      "loss": 0.9638,
+      "step": 2663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064276964996458e-05,
+      "loss": 0.9151,
+      "step": 2664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062653483846223e-05,
+      "loss": 0.9599,
+      "step": 2665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061029395196186e-05,
+      "loss": 0.9814,
+      "step": 2666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805940469916871e-05,
+      "loss": 0.9668,
+      "step": 2667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057779395886216e-05,
+      "loss": 0.9493,
+      "step": 2668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056153485471167e-05,
+      "loss": 0.9474,
+      "step": 2669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054526968046067e-05,
+      "loss": 0.9712,
+      "step": 2670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052899843733475e-05,
+      "loss": 0.9883,
+      "step": 2671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051272112655983e-05,
+      "loss": 0.96,
+      "step": 2672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049643774936246e-05,
+      "loss": 0.9724,
+      "step": 2673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804801483069695e-05,
+      "loss": 0.9718,
+      "step": 2674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046385280060832e-05,
+      "loss": 0.9509,
+      "step": 2675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8044755123150673e-05,
+      "loss": 0.9383,
+      "step": 2676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80431243600893e-05,
+      "loss": 1.0206,
+      "step": 2677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041492990999595e-05,
+      "loss": 0.9355,
+      "step": 2678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8039861016004464e-05,
+      "loss": 0.9972,
+      "step": 2679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038228435226887e-05,
+      "loss": 0.9267,
+      "step": 2680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036595248789862e-05,
+      "loss": 0.9748,
+      "step": 2681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034961456816452e-05,
+      "loss": 0.9731,
+      "step": 2682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033327059429756e-05,
+      "loss": 0.9758,
+      "step": 2683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031692056752926e-05,
+      "loss": 0.9676,
+      "step": 2684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030056448909148e-05,
+      "loss": 0.9209,
+      "step": 2685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028420236021668e-05,
+      "loss": 0.9647,
+      "step": 2686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8026783418213768e-05,
+      "loss": 0.9634,
+      "step": 2687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025145995608775e-05,
+      "loss": 0.9652,
+      "step": 2688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8023507968330068e-05,
+      "loss": 0.9534,
+      "step": 2689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8021869336501066e-05,
+      "loss": 1.0037,
+      "step": 2690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8020230100245237e-05,
+      "loss": 0.8985,
+      "step": 2691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801859025968609e-05,
+      "loss": 1.0356,
+      "step": 2692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016949814947188e-05,
+      "loss": 0.9663,
+      "step": 2693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801530876615213e-05,
+      "loss": 0.9592,
+      "step": 2694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013667113424566e-05,
+      "loss": 0.9622,
+      "step": 2695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801202485688819e-05,
+      "loss": 0.9859,
+      "step": 2696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8010381996666746e-05,
+      "loss": 1.0046,
+      "step": 2697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008738532884012e-05,
+      "loss": 0.925,
+      "step": 2698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007094465663823e-05,
+      "loss": 0.9834,
+      "step": 2699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800544979513005e-05,
+      "loss": 1.0053,
+      "step": 2700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003804521406624e-05,
+      "loss": 0.9798,
+      "step": 2701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8002158644617508e-05,
+      "loss": 0.9564,
+      "step": 2702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800051216488671e-05,
+      "loss": 0.997,
+      "step": 2703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799886508233829e-05,
+      "loss": 0.9832,
+      "step": 2704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997217397096355e-05,
+      "loss": 0.9883,
+      "step": 2705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799556910928505e-05,
+      "loss": 0.9728,
+      "step": 2706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799392021902857e-05,
+      "loss": 0.9818,
+      "step": 2707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992270726451158e-05,
+      "loss": 0.9858,
+      "step": 2708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990620631677092e-05,
+      "loss": 0.9754,
+      "step": 2709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988969934830713e-05,
+      "loss": 0.9652,
+      "step": 2710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987318636036382e-05,
+      "loss": 1.0131,
+      "step": 2711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798566673541853e-05,
+      "loss": 0.9582,
+      "step": 2712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984014233101628e-05,
+      "loss": 0.9478,
+      "step": 2713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7982361129210172e-05,
+      "loss": 0.9985,
+      "step": 2714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980707423868736e-05,
+      "loss": 0.9314,
+      "step": 2715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797905311720191e-05,
+      "loss": 0.9934,
+      "step": 2716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977398209334347e-05,
+      "loss": 0.9623,
+      "step": 2717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797574270039074e-05,
+      "loss": 0.9752,
+      "step": 2718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974086590495825e-05,
+      "loss": 0.9445,
+      "step": 2719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797242987977439e-05,
+      "loss": 0.9338,
+      "step": 2720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7970772568351255e-05,
+      "loss": 0.9056,
+      "step": 2721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969114656351308e-05,
+      "loss": 0.965,
+      "step": 2722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7967456143899453e-05,
+      "loss": 0.9884,
+      "step": 2723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965797031120667e-05,
+      "loss": 0.9552,
+      "step": 2724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796413731813995e-05,
+      "loss": 0.9447,
+      "step": 2725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962477005082367e-05,
+      "loss": 0.9556,
+      "step": 2726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796081609207301e-05,
+      "loss": 1.004,
+      "step": 2727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959154579237027e-05,
+      "loss": 0.946,
+      "step": 2728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7957492466699606e-05,
+      "loss": 0.9339,
+      "step": 2729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795582975458599e-05,
+      "loss": 0.9894,
+      "step": 2730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954166443021456e-05,
+      "loss": 0.9975,
+      "step": 2731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7952502532131326e-05,
+      "loss": 0.9775,
+      "step": 2732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950838022040978e-05,
+      "loss": 0.9901,
+      "step": 2733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949172912875824e-05,
+      "loss": 0.9936,
+      "step": 2734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794750720476133e-05,
+      "loss": 1.0102,
+      "step": 2735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7945840897823e-05,
+      "loss": 0.5273,
+      "step": 2736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794417399218638e-05,
+      "loss": 1.0073,
+      "step": 2737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794250648797708e-05,
+      "loss": 1.0095,
+      "step": 2738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940838385320732e-05,
+      "loss": 0.9994,
+      "step": 2739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793916968434303e-05,
+      "loss": 0.9465,
+      "step": 2740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937500385169695e-05,
+      "loss": 0.9716,
+      "step": 2741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7935830487926516e-05,
+      "loss": 0.8794,
+      "step": 2742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934159992739314e-05,
+      "loss": 0.9768,
+      "step": 2743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793248889973395e-05,
+      "loss": 0.9616,
+      "step": 2744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930817209036338e-05,
+      "loss": 0.932,
+      "step": 2745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792914492077244e-05,
+      "loss": 0.9569,
+      "step": 2746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7927472035068252e-05,
+      "loss": 0.93,
+      "step": 2747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925798552049832e-05,
+      "loss": 0.9643,
+      "step": 2748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924124471843263e-05,
+      "loss": 0.9644,
+      "step": 2749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922449794574686e-05,
+      "loss": 0.9789,
+      "step": 2750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792077452037028e-05,
+      "loss": 0.8998,
+      "step": 2751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791909864935628e-05,
+      "loss": 0.9501,
+      "step": 2752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917422181658954e-05,
+      "loss": 0.9193,
+      "step": 2753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791574511740462e-05,
+      "loss": 0.8959,
+      "step": 2754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914067456719638e-05,
+      "loss": 0.9344,
+      "step": 2755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791238919973042e-05,
+      "loss": 0.943,
+      "step": 2756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910710346563417e-05,
+      "loss": 0.9624,
+      "step": 2757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909030897345118e-05,
+      "loss": 1.0052,
+      "step": 2758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907350852202078e-05,
+      "loss": 0.9757,
+      "step": 2759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905670211260874e-05,
+      "loss": 0.933,
+      "step": 2760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7903988974648145e-05,
+      "loss": 0.9331,
+      "step": 2761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790230714249056e-05,
+      "loss": 0.9952,
+      "step": 2762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790062471491485e-05,
+      "loss": 0.9552,
+      "step": 2763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789894169204778e-05,
+      "loss": 0.9753,
+      "step": 2764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897258074016152e-05,
+      "loss": 0.9705,
+      "step": 2765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789557386094683e-05,
+      "loss": 0.9635,
+      "step": 2766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893889052966713e-05,
+      "loss": 0.9457,
+      "step": 2767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892203650202747e-05,
+      "loss": 0.9576,
+      "step": 2768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890517652781923e-05,
+      "loss": 0.9267,
+      "step": 2769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7888831060831276e-05,
+      "loss": 0.9358,
+      "step": 2770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7887143874477887e-05,
+      "loss": 0.9372,
+      "step": 2771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885456093848882e-05,
+      "loss": 0.9596,
+      "step": 2772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.788376771907143e-05,
+      "loss": 0.9254,
+      "step": 2773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.788207875027274e-05,
+      "loss": 0.9738,
+      "step": 2774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7880389187580084e-05,
+      "loss": 0.9967,
+      "step": 2775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7878699031120756e-05,
+      "loss": 0.969,
+      "step": 2776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7877008281022107e-05,
+      "loss": 0.9422,
+      "step": 2777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787531693741153e-05,
+      "loss": 0.9942,
+      "step": 2778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787362500041647e-05,
+      "loss": 0.9241,
+      "step": 2779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871932470164396e-05,
+      "loss": 0.9484,
+      "step": 2780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787023934678285e-05,
+      "loss": 0.9811,
+      "step": 2781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868545630399395e-05,
+      "loss": 0.9923,
+      "step": 2782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866851321141655e-05,
+      "loss": 0.974,
+      "step": 2783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786515641913729e-05,
+      "loss": 0.9432,
+      "step": 2784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7863460924514e-05,
+      "loss": 0.5199,
+      "step": 2785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7861764837399544e-05,
+      "loss": 0.9459,
+      "step": 2786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786006815792171e-05,
+      "loss": 0.9895,
+      "step": 2787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858370886208353e-05,
+      "loss": 0.9279,
+      "step": 2788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785667302238734e-05,
+      "loss": 0.9619,
+      "step": 2789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785497456658661e-05,
+      "loss": 0.9235,
+      "step": 2790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7853275518934134e-05,
+      "loss": 1.0014,
+      "step": 2791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851575879557937e-05,
+      "loss": 0.9511,
+      "step": 2792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849875648586075e-05,
+      "loss": 0.9599,
+      "step": 2793
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784817482614666e-05,
+      "loss": 0.904,
+      "step": 2794
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846473412367845e-05,
+      "loss": 0.9942,
+      "step": 2795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844771407377822e-05,
+      "loss": 0.5994,
+      "step": 2796
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843068811304838e-05,
+      "loss": 0.9756,
+      "step": 2797
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7841365624277176e-05,
+      "loss": 0.6198,
+      "step": 2798
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783966184642317e-05,
+      "loss": 0.9804,
+      "step": 2799
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783795747787119e-05,
+      "loss": 0.9198,
+      "step": 2800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783625251874966e-05,
+      "loss": 0.9736,
+      "step": 2801
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7834546969187044e-05,
+      "loss": 0.9713,
+      "step": 2802
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783284082931185e-05,
+      "loss": 0.8873,
+      "step": 2803
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831134099252633e-05,
+      "loss": 0.9837,
+      "step": 2804
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829426779137985e-05,
+      "loss": 0.9112,
+      "step": 2805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827718869096554e-05,
+      "loss": 0.5895,
+      "step": 2806
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7826010369257023e-05,
+      "loss": 0.9419,
+      "step": 2807
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824301279748126e-05,
+      "loss": 0.9871,
+      "step": 2808
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822591600698632e-05,
+      "loss": 0.9956,
+      "step": 2809
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7820881332237366e-05,
+      "loss": 0.9536,
+      "step": 2810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7819170474493195e-05,
+      "loss": 0.9503,
+      "step": 2811
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817459027595023e-05,
+      "loss": 0.9462,
+      "step": 2812
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7815746991671804e-05,
+      "loss": 1.0,
+      "step": 2813
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7814034366852537e-05,
+      "loss": 0.9384,
+      "step": 2814
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7812321153266258e-05,
+      "loss": 0.9491,
+      "step": 2815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7810607351042062e-05,
+      "loss": 0.9547,
+      "step": 2816
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7808892960309075e-05,
+      "loss": 0.9642,
+      "step": 2817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780717798119647e-05,
+      "loss": 0.9572,
+      "step": 2818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780546241383347e-05,
+      "loss": 0.9806,
+      "step": 2819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803746258349338e-05,
+      "loss": 0.8998,
+      "step": 2820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7802029514873377e-05,
+      "loss": 0.9384,
+      "step": 2821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800312183534946e-05,
+      "loss": 0.9832,
+      "step": 2822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7798594264463435e-05,
+      "loss": 0.9415,
+      "step": 2823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779687575778829e-05,
+      "loss": 0.9746,
+      "step": 2824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795156663638993e-05,
+      "loss": 0.9522,
+      "step": 2825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793436982145073e-05,
+      "loss": 0.9461,
+      "step": 2826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7791716713436107e-05,
+      "loss": 0.9267,
+      "step": 2827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778999585764171e-05,
+      "loss": 0.933,
+      "step": 2828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788274414891543e-05,
+      "loss": 0.9838,
+      "step": 2829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788274414891543e-05,
+      "loss": 1.0006,
+      "step": 2830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786552385315314e-05,
+      "loss": 0.9523,
+      "step": 2831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778482976904277e-05,
+      "loss": 0.9777,
+      "step": 2832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778310656620371e-05,
+      "loss": 0.9145,
+      "step": 2833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778138277692797e-05,
+      "loss": 0.9624,
+      "step": 2834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779658401345437e-05,
+      "loss": 0.9474,
+      "step": 2835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7777933439586035e-05,
+      "loss": 0.9492,
+      "step": 2836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7776207891779736e-05,
+      "loss": 0.962,
+      "step": 2837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7774481758056553e-05,
+      "loss": 0.9352,
+      "step": 2838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7772755038546552e-05,
+      "loss": 0.9965,
+      "step": 2839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777102773337983e-05,
+      "loss": 0.9495,
+      "step": 2840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769299842686537e-05,
+      "loss": 0.9262,
+      "step": 2841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776757136659687e-05,
+      "loss": 0.9306,
+      "step": 2842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7765842305241055e-05,
+      "loss": 0.9926,
+      "step": 2843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776411265874938e-05,
+      "loss": 0.977,
+      "step": 2844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776238242725217e-05,
+      "loss": 0.9526,
+      "step": 2845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760651610879786e-05,
+      "loss": 0.9948,
+      "step": 2846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758920209762646e-05,
+      "loss": 0.9557,
+      "step": 2847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7757188224031206e-05,
+      "loss": 0.9382,
+      "step": 2848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775545565381597e-05,
+      "loss": 0.9743,
+      "step": 2849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775372249924748e-05,
+      "loss": 0.9469,
+      "step": 2850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751988760456316e-05,
+      "loss": 0.963,
+      "step": 2851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750254437573127e-05,
+      "loss": 0.9823,
+      "step": 2852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748519530728578e-05,
+      "loss": 0.9383,
+      "step": 2853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774678404005339e-05,
+      "loss": 0.928,
+      "step": 2854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774504796567833e-05,
+      "loss": 0.9585,
+      "step": 2855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743311307734212e-05,
+      "loss": 0.9358,
+      "step": 2856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774157406635188e-05,
+      "loss": 0.9298,
+      "step": 2857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739836241662235e-05,
+      "loss": 0.9584,
+      "step": 2858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738097833796218e-05,
+      "loss": 0.934,
+      "step": 2859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773635884288481e-05,
+      "loss": 0.9927,
+      "step": 2860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734619269059043e-05,
+      "loss": 0.9646,
+      "step": 2861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732879112449987e-05,
+      "loss": 0.9366,
+      "step": 2862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731138373188762e-05,
+      "loss": 0.9933,
+      "step": 2863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729397051406523e-05,
+      "loss": 0.9412,
+      "step": 2864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772765514723448e-05,
+      "loss": 0.9974,
+      "step": 2865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772591266080387e-05,
+      "loss": 0.5905,
+      "step": 2866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7724169592245996e-05,
+      "loss": 0.9917,
+      "step": 2867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722425941692193e-05,
+      "loss": 0.9684,
+      "step": 2868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772068170927383e-05,
+      "loss": 0.9858,
+      "step": 2869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7718936895122344e-05,
+      "loss": 0.9719,
+      "step": 2870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7717191499369195e-05,
+      "loss": 0.9658,
+      "step": 2871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715445522145894e-05,
+      "loss": 0.9175,
+      "step": 2872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713698963583994e-05,
+      "loss": 0.964,
+      "step": 2873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77119518238151e-05,
+      "loss": 0.9537,
+      "step": 2874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7710204102970844e-05,
+      "loss": 0.9697,
+      "step": 2875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7708455801182927e-05,
+      "loss": 0.9779,
+      "step": 2876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7706706918583065e-05,
+      "loss": 0.9594,
+      "step": 2877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770495745530304e-05,
+      "loss": 0.9597,
+      "step": 2878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703207411474666e-05,
+      "loss": 0.9859,
+      "step": 2879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7701456787229805e-05,
+      "loss": 0.928,
+      "step": 2880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7699705582700364e-05,
+      "loss": 0.9133,
+      "step": 2881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7697953798018292e-05,
+      "loss": 0.9393,
+      "step": 2882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696201433315572e-05,
+      "loss": 0.9876,
+      "step": 2883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769444848872425e-05,
+      "loss": 0.9816,
+      "step": 2884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7692694964376404e-05,
+      "loss": 1.0319,
+      "step": 2885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690940860404158e-05,
+      "loss": 0.9538,
+      "step": 2886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7689186176939675e-05,
+      "loss": 0.9638,
+      "step": 2887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768743091411517e-05,
+      "loss": 0.9305,
+      "step": 2888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7685675072062894e-05,
+      "loss": 0.9616,
+      "step": 2889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7683918650915148e-05,
+      "loss": 0.9303,
+      "step": 2890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.768216165080427e-05,
+      "loss": 0.9326,
+      "step": 2891
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7680404071862653e-05,
+      "loss": 0.9784,
+      "step": 2892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7678645914222718e-05,
+      "loss": 0.9821,
+      "step": 2893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767688717801694e-05,
+      "loss": 0.9638,
+      "step": 2894
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767512786337784e-05,
+      "loss": 0.9773,
+      "step": 2895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767336797043797e-05,
+      "loss": 1.0104,
+      "step": 2896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767160749932994e-05,
+      "loss": 0.9227,
+      "step": 2897
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669846450186384e-05,
+      "loss": 0.9661,
+      "step": 2898
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668084823140012e-05,
+      "loss": 0.8874,
+      "step": 2899
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7666322618323546e-05,
+      "loss": 0.9319,
+      "step": 2900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7664559835869763e-05,
+      "loss": 0.9423,
+      "step": 2901
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662796475911488e-05,
+      "loss": 0.9662,
+      "step": 2902
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7661032538581586e-05,
+      "loss": 0.9984,
+      "step": 2903
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7659268024012962e-05,
+      "loss": 1.0079,
+      "step": 2904
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657502932338568e-05,
+      "loss": 0.9478,
+      "step": 2905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655737263691398e-05,
+      "loss": 0.5602,
+      "step": 2906
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7653971018204498e-05,
+      "loss": 0.9671,
+      "step": 2907
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765220419601094e-05,
+      "loss": 0.9738,
+      "step": 2908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650436797243854e-05,
+      "loss": 0.9854,
+      "step": 2909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764866882203641e-05,
+      "loss": 0.9391,
+      "step": 2910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646900270521817e-05,
+      "loss": 0.9666,
+      "step": 2911
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645131142833336e-05,
+      "loss": 0.9302,
+      "step": 2912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764336143910426e-05,
+      "loss": 1.0618,
+      "step": 2913
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641591159467934e-05,
+      "loss": 0.9124,
+      "step": 2914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639820304057745e-05,
+      "loss": 0.9247,
+      "step": 2915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638048873007122e-05,
+      "loss": 0.9463,
+      "step": 2916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636276866449535e-05,
+      "loss": 0.9484,
+      "step": 2917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763450428451851e-05,
+      "loss": 0.9546,
+      "step": 2918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632731127347588e-05,
+      "loss": 0.9555,
+      "step": 2919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7630957395070387e-05,
+      "loss": 0.9794,
+      "step": 2920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629183087820546e-05,
+      "loss": 0.9815,
+      "step": 2921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627408205731762e-05,
+      "loss": 0.939,
+      "step": 2922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762563274893776e-05,
+      "loss": 1.0041,
+      "step": 2923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762385671757232e-05,
+      "loss": 0.9834,
+      "step": 2924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622080111769257e-05,
+      "loss": 0.9588,
+      "step": 2925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762030293166244e-05,
+      "loss": 0.9571,
+      "step": 2926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618525177385768e-05,
+      "loss": 0.9697,
+      "step": 2927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7616746849073195e-05,
+      "loss": 1.0105,
+      "step": 2928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761496794685871e-05,
+      "loss": 0.9351,
+      "step": 2929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613188470876356e-05,
+      "loss": 0.968,
+      "step": 2930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76114084212602e-05,
+      "loss": 0.9889,
+      "step": 2931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7609627798144373e-05,
+      "loss": 0.9805,
+      "step": 2932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607846601663037e-05,
+      "loss": 0.9642,
+      "step": 2933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606064831950403e-05,
+      "loss": 0.9396,
+      "step": 2934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7604282489140718e-05,
+      "loss": 0.9511,
+      "step": 2935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760249957336828e-05,
+      "loss": 0.993,
+      "step": 2936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760071608476743e-05,
+      "loss": 0.9639,
+      "step": 2937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598932023472546e-05,
+      "loss": 0.9625,
+      "step": 2938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759714738961805e-05,
+      "loss": 0.9235,
+      "step": 2939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759536218333841e-05,
+      "loss": 0.931,
+      "step": 2940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593576404768146e-05,
+      "loss": 1.0075,
+      "step": 2941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591790054041796e-05,
+      "loss": 0.9997,
+      "step": 2942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590003131293967e-05,
+      "loss": 0.9746,
+      "step": 2943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7588215636659297e-05,
+      "loss": 0.955,
+      "step": 2944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758642757027247e-05,
+      "loss": 0.5573,
+      "step": 2945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758463893226821e-05,
+      "loss": 0.9409,
+      "step": 2946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758284972278129e-05,
+      "loss": 0.9566,
+      "step": 2947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581059941946517e-05,
+      "loss": 0.9187,
+      "step": 2948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757926958989875e-05,
+      "loss": 0.9763,
+      "step": 2949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577478666772886e-05,
+      "loss": 0.9579,
+      "step": 2950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575687172703865e-05,
+      "loss": 0.9558,
+      "step": 2951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757389510782667e-05,
+      "loss": 0.9973,
+      "step": 2952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7572102472276335e-05,
+      "loss": 0.9514,
+      "step": 2953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757030926618793e-05,
+      "loss": 0.9474,
+      "step": 2954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568515489696558e-05,
+      "loss": 0.9945,
+      "step": 2955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7566721142937386e-05,
+      "loss": 1.0042,
+      "step": 2956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564926226045612e-05,
+      "loss": 0.8981,
+      "step": 2957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756313073915647e-05,
+      "loss": 0.9437,
+      "step": 2958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7561334682405257e-05,
+      "loss": 0.5731,
+      "step": 2959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559538055927293e-05,
+      "loss": 0.9331,
+      "step": 2960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7557740859857953e-05,
+      "loss": 0.9309,
+      "step": 2961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7555943094332647e-05,
+      "loss": 0.5477,
+      "step": 2962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7554144759486833e-05,
+      "loss": 0.9673,
+      "step": 2963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7552345855456017e-05,
+      "loss": 0.9836,
+      "step": 2964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7550546382375734e-05,
+      "loss": 0.9783,
+      "step": 2965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754874634038157e-05,
+      "loss": 1.0108,
+      "step": 2966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7546945729609162e-05,
+      "loss": 0.9602,
+      "step": 2967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7545144550194174e-05,
+      "loss": 0.958,
+      "step": 2968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543342802272318e-05,
+      "loss": 0.943,
+      "step": 2969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541540485979357e-05,
+      "loss": 0.955,
+      "step": 2970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539737601451084e-05,
+      "loss": 0.9363,
+      "step": 2971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7537934148823353e-05,
+      "loss": 0.9343,
+      "step": 2972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536130128232035e-05,
+      "loss": 0.6064,
+      "step": 2973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753432553981307e-05,
+      "loss": 0.5822,
+      "step": 2974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7532520383702425e-05,
+      "loss": 0.9559,
+      "step": 2975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7530714660036112e-05,
+      "loss": 0.9656,
+      "step": 2976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528908368950187e-05,
+      "loss": 0.9627,
+      "step": 2977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7527101510580754e-05,
+      "loss": 0.9419,
+      "step": 2978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752529408506395e-05,
+      "loss": 0.5571,
+      "step": 2979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7523486092535962e-05,
+      "loss": 0.9641,
+      "step": 2980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752167753313302e-05,
+      "loss": 0.9869,
+      "step": 2981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751986840699139e-05,
+      "loss": 0.9719,
+      "step": 2982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518058714247383e-05,
+      "loss": 0.948,
+      "step": 2983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516248455037363e-05,
+      "loss": 0.5585,
+      "step": 2984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751443762949772e-05,
+      "loss": 0.9635,
+      "step": 2985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7512626237764903e-05,
+      "loss": 0.6638,
+      "step": 2986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7510814279975387e-05,
+      "loss": 0.9314,
+      "step": 2987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509001756265704e-05,
+      "loss": 0.9619,
+      "step": 2988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750718866677242e-05,
+      "loss": 0.9159,
+      "step": 2989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750537501163215e-05,
+      "loss": 0.9836,
+      "step": 2990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503560790981545e-05,
+      "loss": 0.9461,
+      "step": 2991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501746004957302e-05,
+      "loss": 0.9699,
+      "step": 2992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749993065369616e-05,
+      "loss": 0.9594,
+      "step": 2993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7498114737334902e-05,
+      "loss": 1.0066,
+      "step": 2994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496298256010353e-05,
+      "loss": 0.9864,
+      "step": 2995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749448120985938e-05,
+      "loss": 0.5477,
+      "step": 2996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492663599018893e-05,
+      "loss": 1.0171,
+      "step": 2997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749084542362584e-05,
+      "loss": 0.9593,
+      "step": 2998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7489026683817226e-05,
+      "loss": 0.9922,
+      "step": 2999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7487207379730078e-05,
+      "loss": 0.9148,
+      "step": 3000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748538751150148e-05,
+      "loss": 0.9576,
+      "step": 3001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483567079268555e-05,
+      "loss": 0.9136,
+      "step": 3002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748174608316846e-05,
+      "loss": 0.9365,
+      "step": 3003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7479924523338413e-05,
+      "loss": 1.0278,
+      "step": 3004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747810239991566e-05,
+      "loss": 0.9407,
+      "step": 3005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747627971303749e-05,
+      "loss": 0.9095,
+      "step": 3006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7474456462841245e-05,
+      "loss": 0.9646,
+      "step": 3007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7472632649464295e-05,
+      "loss": 0.9365,
+      "step": 3008
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.747080827304406e-05,
+      "loss": 0.9021,
+      "step": 3009
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7468983333718006e-05,
+      "loss": 0.9108,
+      "step": 3010
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7467157831623636e-05,
+      "loss": 0.9558,
+      "step": 3011
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74653317668985e-05,
+      "loss": 0.9489,
+      "step": 3012
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.746350513968018e-05,
+      "loss": 0.9536,
+      "step": 3013
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7461677950106306e-05,
+      "loss": 0.9522,
+      "step": 3014
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7459850198314562e-05,
+      "loss": 0.9674,
+      "step": 3015
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745802188444266e-05,
+      "loss": 0.9907,
+      "step": 3016
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7456193008628353e-05,
+      "loss": 0.9786,
+      "step": 3017
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454363571009452e-05,
+      "loss": 0.9214,
+      "step": 3018
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452533571723793e-05,
+      "loss": 0.9363,
+      "step": 3019
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450703010909263e-05,
+      "loss": 0.9755,
+      "step": 3020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7448871888703792e-05,
+      "loss": 0.9878,
+      "step": 3021
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744704020524535e-05,
+      "loss": 0.9969,
+      "step": 3022
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744520796067195e-05,
+      "loss": 1.0069,
+      "step": 3023
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744337515512164e-05,
+      "loss": 0.9514,
+      "step": 3024
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7441541788732528e-05,
+      "loss": 0.8625,
+      "step": 3025
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743970786164275e-05,
+      "loss": 0.9452,
+      "step": 3026
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437873373990478e-05,
+      "loss": 0.9593,
+      "step": 3027
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436038325913946e-05,
+      "loss": 0.9779,
+      "step": 3028
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434202717551417e-05,
+      "loss": 0.9725,
+      "step": 3029
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432366549041203e-05,
+      "loss": 0.9775,
+      "step": 3030
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430529820521647e-05,
+      "loss": 0.9675,
+      "step": 3031
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742869253213115e-05,
+      "loss": 0.9732,
+      "step": 3032
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742685468400814e-05,
+      "loss": 0.9607,
+      "step": 3033
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74250162762911e-05,
+      "loss": 0.5237,
+      "step": 3034
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742317730911854e-05,
+      "loss": 0.9834,
+      "step": 3035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742133778262903e-05,
+      "loss": 0.9601,
+      "step": 3036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419497696961172e-05,
+      "loss": 0.9548,
+      "step": 3037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741765705225361e-05,
+      "loss": 0.9817,
+      "step": 3038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415815848645032e-05,
+      "loss": 0.946,
+      "step": 3039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741397408627417e-05,
+      "loss": 0.9422,
+      "step": 3040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741213176527979e-05,
+      "loss": 0.9419,
+      "step": 3041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410288885800716e-05,
+      "loss": 0.9738,
+      "step": 3042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408445447975794e-05,
+      "loss": 0.9589,
+      "step": 3043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740660145194393e-05,
+      "loss": 0.9519,
+      "step": 3044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404756897844054e-05,
+      "loss": 0.939,
+      "step": 3045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740291178581516e-05,
+      "loss": 0.5443,
+      "step": 3046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401066115996265e-05,
+      "loss": 0.9666,
+      "step": 3047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399219888526438e-05,
+      "loss": 0.9414,
+      "step": 3048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739737310354479e-05,
+      "loss": 0.9323,
+      "step": 3049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7395525761190466e-05,
+      "loss": 0.9552,
+      "step": 3050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739367786160266e-05,
+      "loss": 0.9985,
+      "step": 3051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391829404920612e-05,
+      "loss": 0.8973,
+      "step": 3052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738998039128359e-05,
+      "loss": 0.9466,
+      "step": 3053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388130820830914e-05,
+      "loss": 0.9763,
+      "step": 3054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738628069370195e-05,
+      "loss": 0.9238,
+      "step": 3055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7384430010036098e-05,
+      "loss": 1.0084,
+      "step": 3056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73825787699728e-05,
+      "loss": 0.9655,
+      "step": 3057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738072697365154e-05,
+      "loss": 0.9636,
+      "step": 3058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737887462121185e-05,
+      "loss": 0.9327,
+      "step": 3059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73770217127933e-05,
+      "loss": 1.0214,
+      "step": 3060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7375168248535498e-05,
+      "loss": 0.952,
+      "step": 3061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73733142285781e-05,
+      "loss": 0.9853,
+      "step": 3062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371459653060806e-05,
+      "loss": 0.9748,
+      "step": 3063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7369604522123348e-05,
+      "loss": 0.971,
+      "step": 3064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7367748835905504e-05,
+      "loss": 0.9596,
+      "step": 3065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365892594547097e-05,
+      "loss": 0.9801,
+      "step": 3066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736403579818799e-05,
+      "loss": 0.9439,
+      "step": 3067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362178446968093e-05,
+      "loss": 0.9436,
+      "step": 3068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7360320541027342e-05,
+      "loss": 1.0162,
+      "step": 3069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358462080505734e-05,
+      "loss": 0.941,
+      "step": 3070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73566030655433e-05,
+      "loss": 1.0025,
+      "step": 3071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7354743496280103e-05,
+      "loss": 0.8976,
+      "step": 3072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735288337285626e-05,
+      "loss": 0.9822,
+      "step": 3073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735102269541193e-05,
+      "loss": 0.9807,
+      "step": 3074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7349161464087312e-05,
+      "loss": 0.9484,
+      "step": 3075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7347299679022637e-05,
+      "loss": 0.8859,
+      "step": 3076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7345437340358194e-05,
+      "loss": 0.9688,
+      "step": 3077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343574448234294e-05,
+      "loss": 0.9619,
+      "step": 3078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7341711002791316e-05,
+      "loss": 1.0029,
+      "step": 3079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339847004169653e-05,
+      "loss": 0.9983,
+      "step": 3080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7337982452509757e-05,
+      "loss": 0.9671,
+      "step": 3081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733611734795212e-05,
+      "loss": 0.951,
+      "step": 3082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7334251690637272e-05,
+      "loss": 0.9486,
+      "step": 3083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733238548070578e-05,
+      "loss": 0.9583,
+      "step": 3084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 0.9511,
+      "step": 3085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328651403555378e-05,
+      "loss": 0.9667,
+      "step": 3086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7326783536617817e-05,
+      "loss": 0.9548,
+      "step": 3087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732491511762632e-05,
+      "loss": 0.9874,
+      "step": 3088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732304614672167e-05,
+      "loss": 0.975,
+      "step": 3089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732117662404469e-05,
+      "loss": 0.985,
+      "step": 3090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731930654973624e-05,
+      "loss": 0.9152,
+      "step": 3091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317435923937224e-05,
+      "loss": 0.9511,
+      "step": 3092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315564746788592e-05,
+      "loss": 0.9683,
+      "step": 3093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313693018431336e-05,
+      "loss": 0.9254,
+      "step": 3094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731182073900648e-05,
+      "loss": 0.9519,
+      "step": 3095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7309947908655096e-05,
+      "loss": 0.9764,
+      "step": 3096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73080745275183e-05,
+      "loss": 0.944,
+      "step": 3097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730620059573724e-05,
+      "loss": 0.9606,
+      "step": 3098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730432611345312e-05,
+      "loss": 0.9744,
+      "step": 3099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730245108080717e-05,
+      "loss": 0.9575,
+      "step": 3100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7300575497940675e-05,
+      "loss": 0.9828,
+      "step": 3101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298699364994952e-05,
+      "loss": 0.9127,
+      "step": 3102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729682268211136e-05,
+      "loss": 0.9192,
+      "step": 3103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729494544943131e-05,
+      "loss": 1.0004,
+      "step": 3104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729306766709624e-05,
+      "loss": 0.9665,
+      "step": 3105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291189335247636e-05,
+      "loss": 0.9437,
+      "step": 3106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728931045402703e-05,
+      "loss": 0.9582,
+      "step": 3107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7287431023575988e-05,
+      "loss": 0.9034,
+      "step": 3108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7285551044036117e-05,
+      "loss": 0.9728,
+      "step": 3109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7283670515549073e-05,
+      "loss": 1.0072,
+      "step": 3110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728178943825655e-05,
+      "loss": 0.9449,
+      "step": 3111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279907812300276e-05,
+      "loss": 0.9426,
+      "step": 3112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7278025637822035e-05,
+      "loss": 0.9131,
+      "step": 3113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7276142914963635e-05,
+      "loss": 0.6253,
+      "step": 3114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727425964386694e-05,
+      "loss": 0.9148,
+      "step": 3115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7272375824673848e-05,
+      "loss": 0.9541,
+      "step": 3116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72704914575263e-05,
+      "loss": 0.9423,
+      "step": 3117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268606542566276e-05,
+      "loss": 0.9132,
+      "step": 3118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266721079935803e-05,
+      "loss": 0.9524,
+      "step": 3119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264835069776945e-05,
+      "loss": 0.9983,
+      "step": 3120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7262948512231806e-05,
+      "loss": 0.9432,
+      "step": 3121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7261061407442536e-05,
+      "loss": 0.9928,
+      "step": 3122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725917375555132e-05,
+      "loss": 1.0078,
+      "step": 3123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725728555670039e-05,
+      "loss": 0.9497,
+      "step": 3124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255396811032014e-05,
+      "loss": 0.5799,
+      "step": 3125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725350751868851e-05,
+      "loss": 0.9615,
+      "step": 3126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7251617679812224e-05,
+      "loss": 0.974,
+      "step": 3127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7249727294545552e-05,
+      "loss": 1.0013,
+      "step": 3128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7247836363030935e-05,
+      "loss": 0.932,
+      "step": 3129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7245944885410845e-05,
+      "loss": 0.9553,
+      "step": 3130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7244052861827803e-05,
+      "loss": 1.0,
+      "step": 3131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7242160292424362e-05,
+      "loss": 0.9951,
+      "step": 3132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240267177343127e-05,
+      "loss": 1.0148,
+      "step": 3133
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238373516726737e-05,
+      "loss": 0.9779,
+      "step": 3134
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236479310717878e-05,
+      "loss": 0.9844,
+      "step": 3135
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234584559459274e-05,
+      "loss": 0.9639,
+      "step": 3136
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723268926309368e-05,
+      "loss": 0.9752,
+      "step": 3137
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230793421763914e-05,
+      "loss": 0.9707,
+      "step": 3138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7228897035612815e-05,
+      "loss": 0.9871,
+      "step": 3139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227000104783274e-05,
+      "loss": 0.9275,
+      "step": 3140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7225102629418217e-05,
+      "loss": 0.9572,
+      "step": 3141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7223204609660613e-05,
+      "loss": 0.9388,
+      "step": 3142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7221306045653477e-05,
+      "loss": 0.9469,
+      "step": 3143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721940693753986e-05,
+      "loss": 0.9569,
+      "step": 3144
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7217507285462854e-05,
+      "loss": 0.9726,
+      "step": 3145
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721560708956559e-05,
+      "loss": 0.9367,
+      "step": 3146
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7213706349991243e-05,
+      "loss": 0.9849,
+      "step": 3147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721180506688304e-05,
+      "loss": 0.9301,
+      "step": 3148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720990324038422e-05,
+      "loss": 0.9595,
+      "step": 3149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208000870638094e-05,
+      "loss": 0.9706,
+      "step": 3150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7206097957787993e-05,
+      "loss": 0.9956,
+      "step": 3151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72041945019773e-05,
+      "loss": 0.9546,
+      "step": 3152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7202290503349436e-05,
+      "loss": 0.9638,
+      "step": 3153
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200385962047863e-05,
+      "loss": 0.9227,
+      "step": 3154
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719848087821608e-05,
+      "loss": 0.9543,
+      "step": 3155
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719657525199763e-05,
+      "loss": 0.9638,
+      "step": 3156
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71946690835361e-05,
+      "loss": 0.9654,
+      "step": 3157
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719276237297511e-05,
+      "loss": 0.9078,
+      "step": 3158
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190855120458333e-05,
+      "loss": 1.0415,
+      "step": 3159
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718894732612947e-05,
+      "loss": 0.6042,
+      "step": 3160
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7187038990132266e-05,
+      "loss": 0.9526,
+      "step": 3161
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7185130112610518e-05,
+      "loss": 0.9848,
+      "step": 3162
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7183220693708047e-05,
+      "loss": 0.5003,
+      "step": 3163
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7181310733568725e-05,
+      "loss": 0.8895,
+      "step": 3164
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7179400232336462e-05,
+      "loss": 0.9526,
+      "step": 3165
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7177489190155213e-05,
+      "loss": 0.9288,
+      "step": 3166
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175577607168964e-05,
+      "loss": 0.9105,
+      "step": 3167
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173665483521757e-05,
+      "loss": 0.9356,
+      "step": 3168
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171752819357653e-05,
+      "loss": 0.9699,
+      "step": 3169
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7169839614820775e-05,
+      "loss": 0.8883,
+      "step": 3170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167925870055273e-05,
+      "loss": 0.9178,
+      "step": 3171
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166011585205348e-05,
+      "loss": 1.0244,
+      "step": 3172
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7164096760415232e-05,
+      "loss": 1.015,
+      "step": 3173
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7162181395829204e-05,
+      "loss": 0.9152,
+      "step": 3174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716026549159158e-05,
+      "loss": 0.9724,
+      "step": 3175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715834904784672e-05,
+      "loss": 0.9501,
+      "step": 3176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156432064739024e-05,
+      "loss": 0.9519,
+      "step": 3177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7154514542412935e-05,
+      "loss": 0.9972,
+      "step": 3178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152596481012922e-05,
+      "loss": 0.936,
+      "step": 3179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150677880683515e-05,
+      "loss": 0.9865,
+      "step": 3180
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714875874156927e-05,
+      "loss": 0.9424,
+      "step": 3181
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7146839063814795e-05,
+      "loss": 0.9598,
+      "step": 3182
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714491884756473e-05,
+      "loss": 0.9137,
+      "step": 3183
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142998092963756e-05,
+      "loss": 0.9722,
+      "step": 3184
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71410768001566e-05,
+      "loss": 0.9273,
+      "step": 3185
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7139154969288026e-05,
+      "loss": 0.9959,
+      "step": 3186
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7137232600502836e-05,
+      "loss": 0.9319,
+      "step": 3187
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135309693945877e-05,
+      "loss": 0.9163,
+      "step": 3188
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713338624976204e-05,
+      "loss": 0.9285,
+      "step": 3189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131462268096248e-05,
+      "loss": 0.9648,
+      "step": 3190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712953774909346e-05,
+      "loss": 0.9203,
+      "step": 3191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7127612692898695e-05,
+      "loss": 0.9768,
+      "step": 3192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125687099656993e-05,
+      "loss": 0.607,
+      "step": 3193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712376096951345e-05,
+      "loss": 1.0072,
+      "step": 3194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712183430261319e-05,
+      "loss": 0.9735,
+      "step": 3195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7119907099101382e-05,
+      "loss": 0.97,
+      "step": 3196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117979359123237e-05,
+      "loss": 0.9241,
+      "step": 3197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116051082824003e-05,
+      "loss": 0.9149,
+      "step": 3198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7114122270348972e-05,
+      "loss": 0.9511,
+      "step": 3199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112192921843477e-05,
+      "loss": 0.8866,
+      "step": 3200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711026303745289e-05,
+      "loss": 0.9361,
+      "step": 3201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7108332617322617e-05,
+      "loss": 0.9506,
+      "step": 3202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106401661598115e-05,
+      "loss": 0.9961,
+      "step": 3203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710447017042488e-05,
+      "loss": 0.9498,
+      "step": 3204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710253814394843e-05,
+      "loss": 0.9306,
+      "step": 3205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100605582314355e-05,
+      "loss": 0.8889,
+      "step": 3206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709867248566826e-05,
+      "loss": 0.9037,
+      "step": 3207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70967388541558e-05,
+      "loss": 0.9474,
+      "step": 3208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709480468792267e-05,
+      "loss": 0.9541,
+      "step": 3209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70928699871146e-05,
+      "loss": 0.9801,
+      "step": 3210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709093475187737e-05,
+      "loss": 0.928,
+      "step": 3211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088998982356794e-05,
+      "loss": 1.0141,
+      "step": 3212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7087062678698726e-05,
+      "loss": 0.9586,
+      "step": 3213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708512584104906e-05,
+      "loss": 0.9921,
+      "step": 3214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7083188469553737e-05,
+      "loss": 1.0003,
+      "step": 3215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708125056435873e-05,
+      "loss": 0.9265,
+      "step": 3216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707931212561005e-05,
+      "loss": 0.9514,
+      "step": 3217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707737315345376e-05,
+      "loss": 0.9446,
+      "step": 3218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075433648035952e-05,
+      "loss": 0.9655,
+      "step": 3219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073493609502767e-05,
+      "loss": 0.9318,
+      "step": 3220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707155303800038e-05,
+      "loss": 0.9239,
+      "step": 3221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7069611933675006e-05,
+      "loss": 0.9524,
+      "step": 3222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7067670296672904e-05,
+      "loss": 0.9387,
+      "step": 3223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065728127140365e-05,
+      "loss": 0.9311,
+      "step": 3224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706378542522374e-05,
+      "loss": 0.9758,
+      "step": 3225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061842191069398e-05,
+      "loss": 0.906,
+      "step": 3226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7059898424823754e-05,
+      "loss": 0.9708,
+      "step": 3227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057954126633268e-05,
+      "loss": 0.9266,
+      "step": 3228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056009296644445e-05,
+      "loss": 0.9524,
+      "step": 3229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7054063935003813e-05,
+      "loss": 0.9794,
+      "step": 3230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7052118041857954e-05,
+      "loss": 0.9243,
+      "step": 3231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050171617353484e-05,
+      "loss": 0.9525,
+      "step": 3232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048224661637065e-05,
+      "loss": 0.9402,
+      "step": 3233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704627717485539e-05,
+      "loss": 0.9778,
+      "step": 3234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70443291571552e-05,
+      "loss": 0.9282,
+      "step": 3235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7042380608683277e-05,
+      "loss": 0.922,
+      "step": 3236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7040431529586427e-05,
+      "loss": 0.9655,
+      "step": 3237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703848192001152e-05,
+      "loss": 0.9551,
+      "step": 3238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703653178010545e-05,
+      "loss": 0.9932,
+      "step": 3239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034581110015156e-05,
+      "loss": 0.9085,
+      "step": 3240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703262990988761e-05,
+      "loss": 0.944,
+      "step": 3241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7030678179869834e-05,
+      "loss": 0.9386,
+      "step": 3242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702872592010889e-05,
+      "loss": 0.9179,
+      "step": 3243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702677313075187e-05,
+      "loss": 0.9814,
+      "step": 3244
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.702481981194591e-05,
+      "loss": 0.5311,
+      "step": 3245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7022865963838195e-05,
+      "loss": 0.9584,
+      "step": 3246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7020911586575935e-05,
+      "loss": 0.934,
+      "step": 3247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7018956680306387e-05,
+      "loss": 0.9333,
+      "step": 3248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7017001245176857e-05,
+      "loss": 1.0009,
+      "step": 3249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7015045281334677e-05,
+      "loss": 0.8952,
+      "step": 3250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7013088788927218e-05,
+      "loss": 0.943,
+      "step": 3251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011131768101906e-05,
+      "loss": 0.9364,
+      "step": 3252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009174219006188e-05,
+      "loss": 0.9676,
+      "step": 3253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7007216141787568e-05,
+      "loss": 0.9797,
+      "step": 3254
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7005257536593577e-05,
+      "loss": 0.9611,
+      "step": 3255
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7003298403571794e-05,
+      "loss": 1.0202,
+      "step": 3256
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7001338742869836e-05,
+      "loss": 0.9194,
+      "step": 3257
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699937855463535e-05,
+      "loss": 0.9202,
+      "step": 3258
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699741783901604e-05,
+      "loss": 0.9796,
+      "step": 3259
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6995456596159637e-05,
+      "loss": 0.9992,
+      "step": 3260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6993494826213917e-05,
+      "loss": 0.9464,
+      "step": 3261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699153252932669e-05,
+      "loss": 0.895,
+      "step": 3262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6989569705645817e-05,
+      "loss": 0.5692,
+      "step": 3263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6987606355319184e-05,
+      "loss": 0.9331,
+      "step": 3264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698564247849473e-05,
+      "loss": 0.98,
+      "step": 3265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6983678075320424e-05,
+      "loss": 0.9772,
+      "step": 3266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6981713145944284e-05,
+      "loss": 0.9677,
+      "step": 3267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979747690514355e-05,
+      "loss": 0.9793,
+      "step": 3268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977781709178735e-05,
+      "loss": 0.9557,
+      "step": 3269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975815202085556e-05,
+      "loss": 1.0031,
+      "step": 3270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697384816938298e-05,
+      "loss": 0.9703,
+      "step": 3271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697188061121923e-05,
+      "loss": 0.9577,
+      "step": 3272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6969912527742547e-05,
+      "loss": 0.9665,
+      "step": 3273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696794391910123e-05,
+      "loss": 0.9458,
+      "step": 3274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965974785443595e-05,
+      "loss": 0.9654,
+      "step": 3275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696400512691802e-05,
+      "loss": 0.9486,
+      "step": 3276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696203494367292e-05,
+      "loss": 1.0027,
+      "step": 3277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696006423585673e-05,
+      "loss": 0.9808,
+      "step": 3278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6958093003617942e-05,
+      "loss": 0.913,
+      "step": 3279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956121247105087e-05,
+      "loss": 0.9649,
+      "step": 3280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954148966466733e-05,
+      "loss": 0.9657,
+      "step": 3281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695217616185148e-05,
+      "loss": 0.9328,
+      "step": 3282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6950202833407976e-05,
+      "loss": 0.9337,
+      "step": 3283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948228981284906e-05,
+      "loss": 0.9779,
+      "step": 3284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946254605630995e-05,
+      "loss": 0.9908,
+      "step": 3285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944279706595005e-05,
+      "loss": 0.9698,
+      "step": 3286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6942304284325745e-05,
+      "loss": 0.9547,
+      "step": 3287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940328338972053e-05,
+      "loss": 0.9206,
+      "step": 3288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693835187068281e-05,
+      "loss": 0.9985,
+      "step": 3289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693637487960694e-05,
+      "loss": 0.926,
+      "step": 3290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693439736589341e-05,
+      "loss": 0.9483,
+      "step": 3291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693241932969121e-05,
+      "loss": 0.9721,
+      "step": 3292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6930440771149386e-05,
+      "loss": 0.9524,
+      "step": 3293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692846169041702e-05,
+      "loss": 0.9578,
+      "step": 3294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6926482087643223e-05,
+      "loss": 0.9689,
+      "step": 3295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692450196297715e-05,
+      "loss": 0.9384,
+      "step": 3296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6922521316568014e-05,
+      "loss": 0.9267,
+      "step": 3297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6920540148565037e-05,
+      "loss": 0.9846,
+      "step": 3298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69185584591175e-05,
+      "loss": 0.9977,
+      "step": 3299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691657624837472e-05,
+      "loss": 0.9506,
+      "step": 3300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6914593516486047e-05,
+      "loss": 0.9431,
+      "step": 3301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691261026360088e-05,
+      "loss": 0.9294,
+      "step": 3302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691062648986865e-05,
+      "loss": 0.6141,
+      "step": 3303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908642195438823e-05,
+      "loss": 0.9784,
+      "step": 3304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690665738046092e-05,
+      "loss": 0.9654,
+      "step": 3305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6904672045084485e-05,
+      "loss": 0.9313,
+      "step": 3306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690268618945911e-05,
+      "loss": 0.9525,
+      "step": 3307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690069981373442e-05,
+      "loss": 0.9123,
+      "step": 3308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6898712918060093e-05,
+      "loss": 0.9241,
+      "step": 3309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689672550258583e-05,
+      "loss": 0.9692,
+      "step": 3310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894737567461375e-05,
+      "loss": 0.9459,
+      "step": 3311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689274911283652e-05,
+      "loss": 0.8894,
+      "step": 3312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6890760138861085e-05,
+      "loss": 0.9503,
+      "step": 3313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888770645684936e-05,
+      "loss": 0.9334,
+      "step": 3314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886780633457975e-05,
+      "loss": 0.9099,
+      "step": 3315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6884790102330143e-05,
+      "loss": 0.9895,
+      "step": 3316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882799052451425e-05,
+      "loss": 0.8826,
+      "step": 3317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688080748397184e-05,
+      "loss": 0.9494,
+      "step": 3318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6878815397041445e-05,
+      "loss": 0.9651,
+      "step": 3319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687682279181034e-05,
+      "loss": 0.997,
+      "step": 3320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874829668428667e-05,
+      "loss": 0.9751,
+      "step": 3321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6872836027046596e-05,
+      "loss": 0.9727,
+      "step": 3322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6870841867814345e-05,
+      "loss": 0.9548,
+      "step": 3323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686884719088217e-05,
+      "loss": 0.9739,
+      "step": 3324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6866851996400366e-05,
+      "loss": 0.9434,
+      "step": 3325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864856284519263e-05,
+      "loss": 0.9197,
+      "step": 3326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686286005538923e-05,
+      "loss": 0.9578,
+      "step": 3327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6860863309160682e-05,
+      "loss": 0.9646,
+      "step": 3328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858866045984066e-05,
+      "loss": 0.9463,
+      "step": 3329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6856868266009874e-05,
+      "loss": 0.9525,
+      "step": 3330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685486996938863e-05,
+      "loss": 0.9498,
+      "step": 3331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6852871156270904e-05,
+      "loss": 0.9847,
+      "step": 3332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6850871826807297e-05,
+      "loss": 0.9938,
+      "step": 3333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6848871981148452e-05,
+      "loss": 1.0047,
+      "step": 3334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684687161944506e-05,
+      "loss": 0.9347,
+      "step": 3335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684487074184784e-05,
+      "loss": 0.9558,
+      "step": 3336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684286934850755e-05,
+      "loss": 0.9885,
+      "step": 3337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840867439574988e-05,
+      "loss": 0.9805,
+      "step": 3338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838865015200995e-05,
+      "loss": 0.903,
+      "step": 3339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836862075536454e-05,
+      "loss": 0.9502,
+      "step": 3340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683485862073227e-05,
+      "loss": 0.963,
+      "step": 3341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683285465093941e-05,
+      "loss": 0.8932,
+      "step": 3342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683085016630886e-05,
+      "loss": 0.9896,
+      "step": 3343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828845166991655e-05,
+      "loss": 0.874,
+      "step": 3344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6826839653138872e-05,
+      "loss": 0.9094,
+      "step": 3345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682483362490161e-05,
+      "loss": 0.9645,
+      "step": 3346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822827082431022e-05,
+      "loss": 0.9438,
+      "step": 3347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820820025878298e-05,
+      "loss": 0.9783,
+      "step": 3348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6818812455394665e-05,
+      "loss": 0.9322,
+      "step": 3349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6816804371131388e-05,
+      "loss": 0.9451,
+      "step": 3350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814795773239766e-05,
+      "loss": 0.9692,
+      "step": 3351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681278666187115e-05,
+      "loss": 0.5854,
+      "step": 3352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810777037176915e-05,
+      "loss": 0.9629,
+      "step": 3353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680876689930848e-05,
+      "loss": 0.6012,
+      "step": 3354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6806756248417307e-05,
+      "loss": 0.9133,
+      "step": 3355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68047450846549e-05,
+      "loss": 0.9846,
+      "step": 3356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680273340817278e-05,
+      "loss": 0.9346,
+      "step": 3357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680072121912253e-05,
+      "loss": 0.9535,
+      "step": 3358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679870851765576e-05,
+      "loss": 0.9453,
+      "step": 3359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679669530392413e-05,
+      "loss": 0.9575,
+      "step": 3360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679468157807932e-05,
+      "loss": 0.9908,
+      "step": 3361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6792667340273063e-05,
+      "loss": 0.9051,
+      "step": 3362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6790652590657125e-05,
+      "loss": 0.9204,
+      "step": 3363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.678863732938332e-05,
+      "loss": 0.9773,
+      "step": 3364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.678662155660348e-05,
+      "loss": 0.96,
+      "step": 3365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6784605272469502e-05,
+      "loss": 0.956,
+      "step": 3366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6782588477133295e-05,
+      "loss": 0.9504,
+      "step": 3367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6780571170746823e-05,
+      "loss": 0.6512,
+      "step": 3368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6778553353462092e-05,
+      "loss": 0.9643,
+      "step": 3369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677653502543113e-05,
+      "loss": 0.9568,
+      "step": 3370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677451618680602e-05,
+      "loss": 0.9521,
+      "step": 3371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6772496837738866e-05,
+      "loss": 0.9557,
+      "step": 3372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770476978381834e-05,
+      "loss": 1.0073,
+      "step": 3373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6768456608887103e-05,
+      "loss": 0.9659,
+      "step": 3374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766435729406913e-05,
+      "loss": 0.9967,
+      "step": 3375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764414340093527e-05,
+      "loss": 0.9826,
+      "step": 3376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6762392441099247e-05,
+      "loss": 0.9503,
+      "step": 3377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6760370032576424e-05,
+      "loss": 0.9643,
+      "step": 3378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675834711467744e-05,
+      "loss": 0.5794,
+      "step": 3379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756323687554713e-05,
+      "loss": 0.9006,
+      "step": 3380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675429975136071e-05,
+      "loss": 0.8941,
+      "step": 3381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6752275306247924e-05,
+      "loss": 0.944,
+      "step": 3382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750250352368892e-05,
+      "loss": 0.9742,
+      "step": 3383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6748224889876188e-05,
+      "loss": 0.9355,
+      "step": 3384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746198918922426e-05,
+      "loss": 0.9946,
+      "step": 3385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6744172439660262e-05,
+      "loss": 0.8944,
+      "step": 3386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742145452242383e-05,
+      "loss": 1.0072,
+      "step": 3387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6740117956821515e-05,
+      "loss": 0.9296,
+      "step": 3388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673808995355043e-05,
+      "loss": 0.9561,
+      "step": 3389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736061442581922e-05,
+      "loss": 0.9327,
+      "step": 3390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673403242406884e-05,
+      "loss": 0.576,
+      "step": 3391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673200289816407e-05,
+      "loss": 0.9451,
+      "step": 3392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672997286502053e-05,
+      "loss": 0.9755,
+      "step": 3393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672794232479117e-05,
+      "loss": 0.9552,
+      "step": 3394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725911277628996e-05,
+      "loss": 0.9649,
+      "step": 3395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672387972368703e-05,
+      "loss": 0.9597,
+      "step": 3396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672184766311836e-05,
+      "loss": 0.9669,
+      "step": 3397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719815096076084e-05,
+      "loss": 0.8714,
+      "step": 3398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6717782022713353e-05,
+      "loss": 0.5044,
+      "step": 3399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6715748443183353e-05,
+      "loss": 0.9204,
+      "step": 3400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713714357639317e-05,
+      "loss": 0.9573,
+      "step": 3401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67116797662345e-05,
+      "loss": 0.9195,
+      "step": 3402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6709644669122205e-05,
+      "loss": 0.8877,
+      "step": 3403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670760906645577e-05,
+      "loss": 0.921,
+      "step": 3404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6705572958388576e-05,
+      "loss": 0.9729,
+      "step": 3405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703536345074035e-05,
+      "loss": 0.995,
+      "step": 3406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67014992266656e-05,
+      "loss": 0.9684,
+      "step": 3407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699461603316765e-05,
+      "loss": 0.9371,
+      "step": 3408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697423475181062e-05,
+      "loss": 0.9581,
+      "step": 3409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695384842412053e-05,
+      "loss": 0.962,
+      "step": 3410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6693345705163343e-05,
+      "loss": 0.9411,
+      "step": 3411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.8685,
+      "step": 3412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668926591784145e-05,
+      "loss": 0.6467,
+      "step": 3413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687225268075665e-05,
+      "loss": 0.8935,
+      "step": 3414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668518411444498e-05,
+      "loss": 0.9574,
+      "step": 3415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683142457103198e-05,
+      "loss": 0.975,
+      "step": 3416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668110029620415e-05,
+      "loss": 0.924,
+      "step": 3417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679057631901707e-05,
+      "loss": 0.9121,
+      "step": 3418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677014464349775e-05,
+      "loss": 0.9525,
+      "step": 3419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667497079370231e-05,
+      "loss": 0.9307,
+      "step": 3420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6672926620113287e-05,
+      "loss": 0.9424,
+      "step": 3421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667088194373674e-05,
+      "loss": 0.9163,
+      "step": 3422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666883676472672e-05,
+      "loss": 0.9577,
+      "step": 3423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666679108323733e-05,
+      "loss": 0.9074,
+      "step": 3424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6664744899422708e-05,
+      "loss": 0.9725,
+      "step": 3425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666269821343703e-05,
+      "loss": 0.9096,
+      "step": 3426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66606510254345e-05,
+      "loss": 0.9942,
+      "step": 3427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665860333556938e-05,
+      "loss": 0.9251,
+      "step": 3428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6656555143995946e-05,
+      "loss": 0.975,
+      "step": 3429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6654506450868537e-05,
+      "loss": 0.9384,
+      "step": 3430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6652457256341504e-05,
+      "loss": 0.9323,
+      "step": 3431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665040756056926e-05,
+      "loss": 0.9658,
+      "step": 3432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664835736370623e-05,
+      "loss": 0.9537,
+      "step": 3433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6646306665906905e-05,
+      "loss": 0.9751,
+      "step": 3434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6644255467325793e-05,
+      "loss": 0.5604,
+      "step": 3435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6642203768117445e-05,
+      "loss": 0.9818,
+      "step": 3436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6640151568436458e-05,
+      "loss": 0.9626,
+      "step": 3437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638098868437453e-05,
+      "loss": 0.957,
+      "step": 3438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6636045668275098e-05,
+      "loss": 0.9639,
+      "step": 3439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6633991968104095e-05,
+      "loss": 0.9631,
+      "step": 3440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663193776807919e-05,
+      "loss": 1.0232,
+      "step": 3441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6629883068355154e-05,
+      "loss": 0.9459,
+      "step": 3442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627827869086806e-05,
+      "loss": 0.9723,
+      "step": 3443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625772170429005e-05,
+      "loss": 0.953,
+      "step": 3444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662371597253664e-05,
+      "loss": 0.9497,
+      "step": 3445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662165927556463e-05,
+      "loss": 0.9209,
+      "step": 3446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6619602079667956e-05,
+      "loss": 0.9661,
+      "step": 3447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6617544385001616e-05,
+      "loss": 0.9558,
+      "step": 3448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661548619172065e-05,
+      "loss": 0.5252,
+      "step": 3449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6613427499980143e-05,
+      "loss": 0.9706,
+      "step": 3450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661136830993521e-05,
+      "loss": 0.9493,
+      "step": 3451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6609308621741003e-05,
+      "loss": 0.8977,
+      "step": 3452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607248435552714e-05,
+      "loss": 0.9795,
+      "step": 3453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6605187751525577e-05,
+      "loss": 0.9628,
+      "step": 3454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6603126569814854e-05,
+      "loss": 0.9401,
+      "step": 3455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6601064890575852e-05,
+      "loss": 0.9429,
+      "step": 3456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6599002713963916e-05,
+      "loss": 0.9269,
+      "step": 3457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6596940040134416e-05,
+      "loss": 0.9501,
+      "step": 3458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6594876869242785e-05,
+      "loss": 0.9256,
+      "step": 3459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6592813201444462e-05,
+      "loss": 0.9761,
+      "step": 3460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659074903689495e-05,
+      "loss": 0.9267,
+      "step": 3461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588684375749767e-05,
+      "loss": 0.9076,
+      "step": 3462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658661921816449e-05,
+      "loss": 0.5478,
+      "step": 3463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658455356429472e-05,
+      "loss": 0.9618,
+      "step": 3464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582487414296097e-05,
+      "loss": 0.942,
+      "step": 3465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580420768324304e-05,
+      "loss": 0.9671,
+      "step": 3466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6578353626535057e-05,
+      "loss": 0.9233,
+      "step": 3467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65762859890841e-05,
+      "loss": 0.9355,
+      "step": 3468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574217856127236e-05,
+      "loss": 0.9336,
+      "step": 3469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6572149227820286e-05,
+      "loss": 0.9468,
+      "step": 3470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6570080104319122e-05,
+      "loss": 0.9372,
+      "step": 3471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656801048577964e-05,
+      "loss": 0.9842,
+      "step": 3472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565940372357787e-05,
+      "loss": 0.9597,
+      "step": 3473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6563869764209538e-05,
+      "loss": 0.9419,
+      "step": 3474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561798661490904e-05,
+      "loss": 0.966,
+      "step": 3475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559727064357943e-05,
+      "loss": 0.906,
+      "step": 3476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6557654972966743e-05,
+      "loss": 0.9492,
+      "step": 3477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555582387473425e-05,
+      "loss": 0.9422,
+      "step": 3478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655350930803416e-05,
+      "loss": 0.9112,
+      "step": 3479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655143573480515e-05,
+      "loss": 0.9496,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6549361667942628e-05,
+      "loss": 0.5709,
+      "step": 3481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.654728710760287e-05,
+      "loss": 1.0079,
+      "step": 3482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.654521205394219e-05,
+      "loss": 0.9639,
+      "step": 3483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6543136507116938e-05,
+      "loss": 0.965,
+      "step": 3484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6541060467283503e-05,
+      "loss": 0.9313,
+      "step": 3485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6538983934598304e-05,
+      "loss": 0.9713,
+      "step": 3486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653690690921781e-05,
+      "loss": 0.9383,
+      "step": 3487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653482939129851e-05,
+      "loss": 0.5449,
+      "step": 3488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653275138099695e-05,
+      "loss": 0.9593,
+      "step": 3489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653067287846969e-05,
+      "loss": 0.939,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652859388387335e-05,
+      "loss": 0.9387,
+      "step": 3491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6526514397364575e-05,
+      "loss": 0.9324,
+      "step": 3492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524434419100047e-05,
+      "loss": 0.5047,
+      "step": 3493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522353949236486e-05,
+      "loss": 0.9758,
+      "step": 3494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6520272987930652e-05,
+      "loss": 0.9543,
+      "step": 3495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651819153533934e-05,
+      "loss": 0.5361,
+      "step": 3496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651610959161938e-05,
+      "loss": 0.9281,
+      "step": 3497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6514027156927645e-05,
+      "loss": 0.9766,
+      "step": 3498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651194423142103e-05,
+      "loss": 0.9482,
+      "step": 3499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509860815256496e-05,
+      "loss": 0.9247,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6507776908591008e-05,
+      "loss": 1.014,
+      "step": 3501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650569251158159e-05,
+      "loss": 0.9662,
+      "step": 3502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503607624385294e-05,
+      "loss": 0.9587,
+      "step": 3503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650152224715921e-05,
+      "loss": 0.6013,
+      "step": 3504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499436380060466e-05,
+      "loss": 0.9368,
+      "step": 3505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497350023246225e-05,
+      "loss": 0.569,
+      "step": 3506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495263176873693e-05,
+      "loss": 0.9484,
+      "step": 3507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493175841100106e-05,
+      "loss": 0.535,
+      "step": 3508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649108801608274e-05,
+      "loss": 0.9309,
+      "step": 3509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6488999701978905e-05,
+      "loss": 0.9751,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648691089894595e-05,
+      "loss": 0.5851,
+      "step": 3511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6484821607141266e-05,
+      "loss": 0.9511,
+      "step": 3512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482731826722268e-05,
+      "loss": 0.9341,
+      "step": 3513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6480641557846424e-05,
+      "loss": 0.5619,
+      "step": 3514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6478550800671224e-05,
+      "loss": 0.948,
+      "step": 3515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64764595553542e-05,
+      "loss": 0.9315,
+      "step": 3516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6474367822052925e-05,
+      "loss": 0.9594,
+      "step": 3517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6472275600925003e-05,
+      "loss": 0.9513,
+      "step": 3518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6470182892128085e-05,
+      "loss": 0.945,
+      "step": 3519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6468089695819845e-05,
+      "loss": 0.9527,
+      "step": 3520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6465996012157996e-05,
+      "loss": 0.5645,
+      "step": 3521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64639018413003e-05,
+      "loss": 0.9254,
+      "step": 3522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461807183404545e-05,
+      "loss": 0.9486,
+      "step": 3523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6459712038628552e-05,
+      "loss": 0.9381,
+      "step": 3524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645761640713019e-05,
+      "loss": 0.9069,
+      "step": 3525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645552028906736e-05,
+      "loss": 0.9812,
+      "step": 3526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453423684597995e-05,
+      "loss": 0.9248,
+      "step": 3527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6451326593880072e-05,
+      "loss": 0.9629,
+      "step": 3528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6449229017071604e-05,
+      "loss": 0.8896,
+      "step": 3529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644713095433063e-05,
+      "loss": 0.9264,
+      "step": 3530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644503240581524e-05,
+      "loss": 0.8984,
+      "step": 3531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6442933371683548e-05,
+      "loss": 0.9053,
+      "step": 3532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644083385209372e-05,
+      "loss": 0.9373,
+      "step": 3533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643873384720394e-05,
+      "loss": 0.9522,
+      "step": 3534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6436633357172443e-05,
+      "loss": 0.9394,
+      "step": 3535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6434532382157495e-05,
+      "loss": 0.518,
+      "step": 3536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432430922317396e-05,
+      "loss": 0.9181,
+      "step": 3537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643032897781049e-05,
+      "loss": 0.9673,
+      "step": 3538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642822654879515e-05,
+      "loss": 0.9523,
+      "step": 3539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6426123635429787e-05,
+      "loss": 0.9708,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6424020237872854e-05,
+      "loss": 0.9303,
+      "step": 3541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421916356282834e-05,
+      "loss": 0.962,
+      "step": 3542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6419811990818252e-05,
+      "loss": 0.976,
+      "step": 3543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641770714163766e-05,
+      "loss": 0.9236,
+      "step": 3544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415601808899658e-05,
+      "loss": 0.941,
+      "step": 3545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641349599276288e-05,
+      "loss": 0.9752,
+      "step": 3546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411389693385985e-05,
+      "loss": 0.9436,
+      "step": 3547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409282910927685e-05,
+      "loss": 0.9786,
+      "step": 3548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640717564554672e-05,
+      "loss": 0.9841,
+      "step": 3549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640506789740186e-05,
+      "loss": 0.9829,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6402959666651924e-05,
+      "loss": 0.9946,
+      "step": 3551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640085095345576e-05,
+      "loss": 0.9606,
+      "step": 3552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398741757972256e-05,
+      "loss": 0.607,
+      "step": 3553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396632080360332e-05,
+      "loss": 0.9241,
+      "step": 3554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639452192077895e-05,
+      "loss": 0.9223,
+      "step": 3555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392411279387098e-05,
+      "loss": 0.9375,
+      "step": 3556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390300156343815e-05,
+      "loss": 0.932,
+      "step": 3557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6388188551808166e-05,
+      "loss": 0.9561,
+      "step": 3558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638607646593925e-05,
+      "loss": 0.9467,
+      "step": 3559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6383963898896215e-05,
+      "loss": 0.9846,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6381850850838232e-05,
+      "loss": 0.9823,
+      "step": 3561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6379737321924517e-05,
+      "loss": 0.5746,
+      "step": 3562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6377623312314316e-05,
+      "loss": 0.8895,
+      "step": 3563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6375508822166917e-05,
+      "loss": 0.9811,
+      "step": 3564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6373393851641642e-05,
+      "loss": 0.9638,
+      "step": 3565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637127840089784e-05,
+      "loss": 0.8837,
+      "step": 3566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6369162470094915e-05,
+      "loss": 0.9354,
+      "step": 3567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636704605939229e-05,
+      "loss": 0.9395,
+      "step": 3568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6364929168949438e-05,
+      "loss": 0.9443,
+      "step": 3569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6362811798925852e-05,
+      "loss": 0.9236,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6360693949481076e-05,
+      "loss": 0.9906,
+      "step": 3571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6358575620774687e-05,
+      "loss": 0.9265,
+      "step": 3572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635645681296629e-05,
+      "loss": 0.8894,
+      "step": 3573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635433752621553e-05,
+      "loss": 0.911,
+      "step": 3574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6352217760682098e-05,
+      "loss": 0.5583,
+      "step": 3575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6350097516525705e-05,
+      "loss": 0.9211,
+      "step": 3576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6347976793906113e-05,
+      "loss": 0.886,
+      "step": 3577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6345855592983107e-05,
+      "loss": 0.9596,
+      "step": 3578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343733913916516e-05,
+      "loss": 0.9482,
+      "step": 3579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63416117568662e-05,
+      "loss": 0.9228,
+      "step": 3580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6339489121992062e-05,
+      "loss": 0.9504,
+      "step": 3581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6337366009454037e-05,
+      "loss": 0.8931,
+      "step": 3582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6335242419412095e-05,
+      "loss": 0.9792,
+      "step": 3583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633311835202624e-05,
+      "loss": 0.9346,
+      "step": 3584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633099380745652e-05,
+      "loss": 0.9257,
+      "step": 3585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632886878586301e-05,
+      "loss": 0.9443,
+      "step": 3586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6326743287405826e-05,
+      "loss": 0.9132,
+      "step": 3587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324617312245123e-05,
+      "loss": 0.8624,
+      "step": 3588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6322490860541077e-05,
+      "loss": 0.9151,
+      "step": 3589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6320363932453922e-05,
+      "loss": 0.9468,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631823652814391e-05,
+      "loss": 0.95,
+      "step": 3591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6316108647771333e-05,
+      "loss": 0.9663,
+      "step": 3592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6313980291496528e-05,
+      "loss": 0.9793,
+      "step": 3593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631185145947986e-05,
+      "loss": 0.9,
+      "step": 3594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630972215188173e-05,
+      "loss": 0.9505,
+      "step": 3595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630759236886257e-05,
+      "loss": 0.8901,
+      "step": 3596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305462110582863e-05,
+      "loss": 0.9618,
+      "step": 3597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6303331377203113e-05,
+      "loss": 0.6113,
+      "step": 3598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6301200168883868e-05,
+      "loss": 0.9937,
+      "step": 3599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62990684857857e-05,
+      "loss": 0.9388,
+      "step": 3600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.629693632806924e-05,
+      "loss": 0.947,
+      "step": 3601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6294803695895127e-05,
+      "loss": 0.9678,
+      "step": 3602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6292670589424057e-05,
+      "loss": 0.9451,
+      "step": 3603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6290537008816753e-05,
+      "loss": 0.9271,
+      "step": 3604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6288402954233976e-05,
+      "loss": 0.884,
+      "step": 3605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6286268425836517e-05,
+      "loss": 0.9263,
+      "step": 3606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6284133423785213e-05,
+      "loss": 0.5868,
+      "step": 3607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6281997948240926e-05,
+      "loss": 0.9224,
+      "step": 3608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627986199936456e-05,
+      "loss": 0.91,
+      "step": 3609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627772557731705e-05,
+      "loss": 0.9708,
+      "step": 3610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275588682259375e-05,
+      "loss": 0.9304,
+      "step": 3611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627345131435254e-05,
+      "loss": 0.9545,
+      "step": 3612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6271313473757596e-05,
+      "loss": 0.9997,
+      "step": 3613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626917516063562e-05,
+      "loss": 0.9458,
+      "step": 3614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6267036375147728e-05,
+      "loss": 0.9509,
+      "step": 3615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626489711745507e-05,
+      "loss": 0.8921,
+      "step": 3616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6262757387718834e-05,
+      "loss": 0.9294,
+      "step": 3617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626061718610025e-05,
+      "loss": 0.9633,
+      "step": 3618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625847651276057e-05,
+      "loss": 0.9369,
+      "step": 3619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6256335367861088e-05,
+      "loss": 0.9354,
+      "step": 3620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6254193751563137e-05,
+      "loss": 1.0018,
+      "step": 3621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6252051664028076e-05,
+      "loss": 0.9209,
+      "step": 3622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6249909105417316e-05,
+      "loss": 0.9582,
+      "step": 3623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6247766075892283e-05,
+      "loss": 0.9291,
+      "step": 3624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6245622575614452e-05,
+      "loss": 0.9601,
+      "step": 3625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6243478604745333e-05,
+      "loss": 0.9756,
+      "step": 3626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6241334163446465e-05,
+      "loss": 0.9754,
+      "step": 3627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623918925187943e-05,
+      "loss": 0.9086,
+      "step": 3628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623704387020584e-05,
+      "loss": 0.9577,
+      "step": 3629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.9446,
+      "step": 3630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6232751697185617e-05,
+      "loss": 0.9702,
+      "step": 3631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6230604906162387e-05,
+      "loss": 0.976,
+      "step": 3632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228457645679414e-05,
+      "loss": 0.9514,
+      "step": 3633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6226309915898483e-05,
+      "loss": 0.9847,
+      "step": 3634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622416171698142e-05,
+      "loss": 0.8935,
+      "step": 3635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6222013049090086e-05,
+      "loss": 0.5312,
+      "step": 3636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6219863912386376e-05,
+      "loss": 0.9507,
+      "step": 3637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621771430703223e-05,
+      "loss": 0.9388,
+      "step": 3638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215564233189606e-05,
+      "loss": 1.0109,
+      "step": 3639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213413691020505e-05,
+      "loss": 0.9698,
+      "step": 3640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621126268068697e-05,
+      "loss": 0.9522,
+      "step": 3641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6209111202351076e-05,
+      "loss": 0.9623,
+      "step": 3642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6206959256174927e-05,
+      "loss": 0.9547,
+      "step": 3643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6204806842320667e-05,
+      "loss": 0.9431,
+      "step": 3644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6202653960950474e-05,
+      "loss": 0.9237,
+      "step": 3645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200500612226566e-05,
+      "loss": 0.9384,
+      "step": 3646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6198346796311188e-05,
+      "loss": 0.9291,
+      "step": 3647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619619251336663e-05,
+      "loss": 0.9515,
+      "step": 3648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194037763555203e-05,
+      "loss": 0.8881,
+      "step": 3649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619188254703927e-05,
+      "loss": 0.951,
+      "step": 3650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189726863981212e-05,
+      "loss": 0.9422,
+      "step": 3651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618757071454346e-05,
+      "loss": 0.9317,
+      "step": 3652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185414098888475e-05,
+      "loss": 0.9529,
+      "step": 3653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6183257017178754e-05,
+      "loss": 0.9361,
+      "step": 3654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618109946957682e-05,
+      "loss": 0.5806,
+      "step": 3655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617894145624524e-05,
+      "loss": 0.9527,
+      "step": 3656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6176782977346626e-05,
+      "loss": 0.9322,
+      "step": 3657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61746240330436e-05,
+      "loss": 0.9069,
+      "step": 3658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617246462349884e-05,
+      "loss": 0.9409,
+      "step": 3659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617030474887505e-05,
+      "loss": 0.923,
+      "step": 3660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6168144409334966e-05,
+      "loss": 0.9419,
+      "step": 3661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6165983605041375e-05,
+      "loss": 1.0158,
+      "step": 3662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163822336157076e-05,
+      "loss": 0.8994,
+      "step": 3663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161660602844924e-05,
+      "loss": 0.968,
+      "step": 3664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159498405267794e-05,
+      "loss": 0.9687,
+      "step": 3665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6157335743588606e-05,
+      "loss": 0.933,
+      "step": 3666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615517261797031e-05,
+      "loss": 0.988,
+      "step": 3667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153009028575894e-05,
+      "loss": 0.9489,
+      "step": 3668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615084497556837e-05,
+      "loss": 0.9476,
+      "step": 3669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148680459110803e-05,
+      "loss": 0.8988,
+      "step": 3670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146515479366282e-05,
+      "loss": 0.9439,
+      "step": 3671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6144350036497925e-05,
+      "loss": 0.9402,
+      "step": 3672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6142184130668902e-05,
+      "loss": 0.9278,
+      "step": 3673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61400177620424e-05,
+      "loss": 0.951,
+      "step": 3674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613785093078166e-05,
+      "loss": 0.9609,
+      "step": 3675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6135683637049938e-05,
+      "loss": 0.958,
+      "step": 3676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6133515881010537e-05,
+      "loss": 0.9241,
+      "step": 3677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131347662826793e-05,
+      "loss": 0.9518,
+      "step": 3678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129178982662073e-05,
+      "loss": 0.8955,
+      "step": 3679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612700984067978e-05,
+      "loss": 0.9589,
+      "step": 3680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612484023704336e-05,
+      "loss": 0.9645,
+      "step": 3681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612267017191628e-05,
+      "loss": 0.9232,
+      "step": 3682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612049964546205e-05,
+      "loss": 0.9274,
+      "step": 3683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118328657844216e-05,
+      "loss": 0.9857,
+      "step": 3684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116157209226356e-05,
+      "loss": 0.9448,
+      "step": 3685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6113985299772083e-05,
+      "loss": 0.9384,
+      "step": 3686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611181292964504e-05,
+      "loss": 0.9472,
+      "step": 3687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6109640099008914e-05,
+      "loss": 0.9138,
+      "step": 3688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610746680802742e-05,
+      "loss": 0.9786,
+      "step": 3689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105293056864314e-05,
+      "loss": 0.9065,
+      "step": 3690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610311884568338e-05,
+      "loss": 0.9003,
+      "step": 3691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6100944174648433e-05,
+      "loss": 0.9398,
+      "step": 3692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6098769043923338e-05,
+      "loss": 0.9817,
+      "step": 3693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096593453671983e-05,
+      "loss": 0.9723,
+      "step": 3694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609441740405829e-05,
+      "loss": 0.9582,
+      "step": 3695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609224089524622e-05,
+      "loss": 0.5508,
+      "step": 3696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6090063927399764e-05,
+      "loss": 0.9587,
+      "step": 3697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6087886500682958e-05,
+      "loss": 0.9702,
+      "step": 3698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6085708615259863e-05,
+      "loss": 0.9651,
+      "step": 3699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608353027129457e-05,
+      "loss": 0.8996,
+      "step": 3700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6081351468951224e-05,
+      "loss": 0.6128,
+      "step": 3701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6079172208393985e-05,
+      "loss": 1.012,
+      "step": 3702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607699248978705e-05,
+      "loss": 0.9248,
+      "step": 3703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074812313294665e-05,
+      "loss": 0.9309,
+      "step": 3704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607263167908109e-05,
+      "loss": 0.951,
+      "step": 3705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607045058731064e-05,
+      "loss": 0.9101,
+      "step": 3706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6068269038147645e-05,
+      "loss": 0.949,
+      "step": 3707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6066087031756485e-05,
+      "loss": 0.9724,
+      "step": 3708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6063904568301572e-05,
+      "loss": 1.0013,
+      "step": 3709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6061721647947343e-05,
+      "loss": 0.5313,
+      "step": 3710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059538270858275e-05,
+      "loss": 0.9603,
+      "step": 3711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057354437198882e-05,
+      "loss": 0.939,
+      "step": 3712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605517014713371e-05,
+      "loss": 0.9279,
+      "step": 3713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605298540082734e-05,
+      "loss": 0.954,
+      "step": 3714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050800198444385e-05,
+      "loss": 0.9525,
+      "step": 3715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6048614540149494e-05,
+      "loss": 0.9022,
+      "step": 3716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6046428426107354e-05,
+      "loss": 0.948,
+      "step": 3717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6044241856482677e-05,
+      "loss": 0.9634,
+      "step": 3718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6042054831440227e-05,
+      "loss": 0.9252,
+      "step": 3719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6039867351144778e-05,
+      "loss": 0.8525,
+      "step": 3720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6037679415761155e-05,
+      "loss": 0.9174,
+      "step": 3721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6035491025454215e-05,
+      "loss": 0.946,
+      "step": 3722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603330218038885e-05,
+      "loss": 0.9122,
+      "step": 3723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031112880729974e-05,
+      "loss": 0.9308,
+      "step": 3724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6028923126642553e-05,
+      "loss": 0.9283,
+      "step": 3725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6026732918291577e-05,
+      "loss": 0.9629,
+      "step": 3726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024542255842077e-05,
+      "loss": 0.8855,
+      "step": 3727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6022351139459106e-05,
+      "loss": 0.9048,
+      "step": 3728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602015956930776e-05,
+      "loss": 0.9401,
+      "step": 3729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6017967545553174e-05,
+      "loss": 0.9619,
+      "step": 3730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015775068360507e-05,
+      "loss": 0.9031,
+      "step": 3731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013582137894957e-05,
+      "loss": 0.9516,
+      "step": 3732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6011388754321757e-05,
+      "loss": 0.9448,
+      "step": 3733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600919491780617e-05,
+      "loss": 0.9648,
+      "step": 3734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6007000628513498e-05,
+      "loss": 0.9694,
+      "step": 3735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6004805886609074e-05,
+      "loss": 0.8871,
+      "step": 3736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6002610692258265e-05,
+      "loss": 0.945,
+      "step": 3737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000415045626474e-05,
+      "loss": 0.8843,
+      "step": 3738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599821894687914e-05,
+      "loss": 0.5574,
+      "step": 3739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599602239618173e-05,
+      "loss": 0.9183,
+      "step": 3740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5993825393699746e-05,
+      "loss": 0.9085,
+      "step": 3741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5991627939598732e-05,
+      "loss": 0.893,
+      "step": 3742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598943003404426e-05,
+      "loss": 0.9517,
+      "step": 3743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598723167720193e-05,
+      "loss": 0.9649,
+      "step": 3744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5985032869237392e-05,
+      "loss": 0.927,
+      "step": 3745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598283361031631e-05,
+      "loss": 0.5419,
+      "step": 3746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59806339006044e-05,
+      "loss": 1.0024,
+      "step": 3747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978433740267402e-05,
+      "loss": 0.9395,
+      "step": 3748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597623312947109e-05,
+      "loss": 0.9429,
+      "step": 3749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597403206838128e-05,
+      "loss": 0.97,
+      "step": 3750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5971830557163813e-05,
+      "loss": 0.9576,
+      "step": 3751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5969628595984566e-05,
+      "loss": 0.9354,
+      "step": 3752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5967426185009448e-05,
+      "loss": 0.9445,
+      "step": 3753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5965223324404414e-05,
+      "loss": 0.8945,
+      "step": 3754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5963020014335437e-05,
+      "loss": 0.9694,
+      "step": 3755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596081625496853e-05,
+      "loss": 0.9466,
+      "step": 3756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5958612046469745e-05,
+      "loss": 0.965,
+      "step": 3757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595640738900516e-05,
+      "loss": 0.9427,
+      "step": 3758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595420228274089e-05,
+      "loss": 0.5634,
+      "step": 3759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5951996727843087e-05,
+      "loss": 0.9845,
+      "step": 3760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949790724477932e-05,
+      "loss": 0.9434,
+      "step": 3761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594758427281164e-05,
+      "loss": 0.9089,
+      "step": 3762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945377373010462e-05,
+      "loss": 0.9768,
+      "step": 3763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943170025240685e-05,
+      "loss": 0.9072,
+      "step": 3764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940962229668625e-05,
+      "loss": 0.8856,
+      "step": 3765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938753986460634e-05,
+      "loss": 0.578,
+      "step": 3766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5936545295783094e-05,
+      "loss": 0.9566,
+      "step": 3767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5934336157802427e-05,
+      "loss": 0.9129,
+      "step": 3768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593212657268509e-05,
+      "loss": 0.8851,
+      "step": 3769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5929916540597557e-05,
+      "loss": 0.9448,
+      "step": 3770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927706061706363e-05,
+      "loss": 0.9646,
+      "step": 3771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5925495136178054e-05,
+      "loss": 0.932,
+      "step": 3772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5923283764179214e-05,
+      "loss": 0.9366,
+      "step": 3773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592107194587647e-05,
+      "loss": 0.9625,
+      "step": 3774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591885968143648e-05,
+      "loss": 0.9301,
+      "step": 3775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916646971025925e-05,
+      "loss": 0.9066,
+      "step": 3776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914433814811524e-05,
+      "loss": 0.8954,
+      "step": 3777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5912220212960042e-05,
+      "loss": 0.9253,
+      "step": 3778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910006165638267e-05,
+      "loss": 0.9828,
+      "step": 3779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5907791673013016e-05,
+      "loss": 0.9791,
+      "step": 3780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905576735251148e-05,
+      "loss": 0.9426,
+      "step": 3781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5903361352519552e-05,
+      "loss": 0.9756,
+      "step": 3782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901145524985155e-05,
+      "loss": 0.9433,
+      "step": 3783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5898929252814907e-05,
+      "loss": 0.9746,
+      "step": 3784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58967125361758e-05,
+      "loss": 0.9704,
+      "step": 3785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894495375234865e-05,
+      "loss": 0.92,
+      "step": 3786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589227777015915e-05,
+      "loss": 0.8966,
+      "step": 3787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5890059721115753e-05,
+      "loss": 0.5573,
+      "step": 3788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5887841228271794e-05,
+      "loss": 0.8746,
+      "step": 3789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588562229179443e-05,
+      "loss": 0.9607,
+      "step": 3790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5883402911850856e-05,
+      "loss": 0.9488,
+      "step": 3791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881183088608293e-05,
+      "loss": 0.9693,
+      "step": 3792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5878962822234002e-05,
+      "loss": 0.9121,
+      "step": 3793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587674211289527e-05,
+      "loss": 0.8746,
+      "step": 3794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5874520960759423e-05,
+      "loss": 0.9806,
+      "step": 3795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587229936599382e-05,
+      "loss": 0.946,
+      "step": 3796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5870077328765857e-05,
+      "loss": 0.9853,
+      "step": 3797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586785484924295e-05,
+      "loss": 0.941,
+      "step": 3798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586563192759256e-05,
+      "loss": 0.9489,
+      "step": 3799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586340856398218e-05,
+      "loss": 0.9096,
+      "step": 3800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5861184758579332e-05,
+      "loss": 0.9612,
+      "step": 3801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5858960511551578e-05,
+      "loss": 0.9083,
+      "step": 3802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585673582306651e-05,
+      "loss": 0.9695,
+      "step": 3803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5854510693291744e-05,
+      "loss": 0.9958,
+      "step": 3804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5852285122394943e-05,
+      "loss": 0.9462,
+      "step": 3805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58500591105438e-05,
+      "loss": 0.9343,
+      "step": 3806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584783265790604e-05,
+      "loss": 0.8956,
+      "step": 3807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5845605764649415e-05,
+      "loss": 0.9268,
+      "step": 3808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5843378430941715e-05,
+      "loss": 0.9037,
+      "step": 3809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584115065695077e-05,
+      "loss": 0.9649,
+      "step": 3810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5838922442844436e-05,
+      "loss": 0.9547,
+      "step": 3811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836693788790597e-05,
+      "loss": 0.9589,
+      "step": 3812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583446469495718e-05,
+      "loss": 0.5193,
+      "step": 3813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5832235161512144e-05,
+      "loss": 0.9087,
+      "step": 3814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5830005188623476e-05,
+      "loss": 0.9393,
+      "step": 3815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5827774776459195e-05,
+      "loss": 0.9077,
+      "step": 3816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5825543925187363e-05,
+      "loss": 0.94,
+      "step": 3817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823312634976068e-05,
+      "loss": 0.962,
+      "step": 3818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5821080905993425e-05,
+      "loss": 0.9061,
+      "step": 3819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58188487384076e-05,
+      "loss": 0.8783,
+      "step": 3820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581661613238677e-05,
+      "loss": 0.9277,
+      "step": 3821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814383088099164e-05,
+      "loss": 0.8791,
+      "step": 3822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812149605713028e-05,
+      "loss": 0.9202,
+      "step": 3823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580991568539666e-05,
+      "loss": 0.9324,
+      "step": 3824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807681327318372e-05,
+      "loss": 0.9466,
+      "step": 3825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805446531646516e-05,
+      "loss": 0.9236,
+      "step": 3826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5803211298549483e-05,
+      "loss": 0.8712,
+      "step": 3827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800975628195692e-05,
+      "loss": 0.9432,
+      "step": 3828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579873952075359e-05,
+      "loss": 0.9793,
+      "step": 3829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579650297639167e-05,
+      "loss": 0.9302,
+      "step": 3830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5794265995278438e-05,
+      "loss": 0.9385,
+      "step": 3831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5792028577582453e-05,
+      "loss": 0.9437,
+      "step": 3832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57897907234723e-05,
+      "loss": 0.9102,
+      "step": 3833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5787552433116586e-05,
+      "loss": 0.9593,
+      "step": 3834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.578531370668397e-05,
+      "loss": 0.9195,
+      "step": 3835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5783074544343132e-05,
+      "loss": 0.93,
+      "step": 3836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5780834946262782e-05,
+      "loss": 0.9217,
+      "step": 3837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5778594912611677e-05,
+      "loss": 0.9448,
+      "step": 3838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5778594912611677e-05,
+      "loss": 0.9293,
+      "step": 3839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577635444355859e-05,
+      "loss": 0.9029,
+      "step": 3840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5774113539272332e-05,
+      "loss": 0.944,
+      "step": 3841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5771872199921755e-05,
+      "loss": 0.9581,
+      "step": 3842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576963042567574e-05,
+      "loss": 0.8908,
+      "step": 3843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5767388216703196e-05,
+      "loss": 0.9671,
+      "step": 3844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5765145573173066e-05,
+      "loss": 0.9731,
+      "step": 3845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576290249525433e-05,
+      "loss": 0.938,
+      "step": 3846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576065898311599e-05,
+      "loss": 0.9542,
+      "step": 3847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5758415036927105e-05,
+      "loss": 0.8781,
+      "step": 3848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575617065685674e-05,
+      "loss": 0.9459,
+      "step": 3849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5753925843073997e-05,
+      "loss": 0.9839,
+      "step": 3850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575168059574803e-05,
+      "loss": 0.9754,
+      "step": 3851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5749434915048004e-05,
+      "loss": 0.9517,
+      "step": 3852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747188801143126e-05,
+      "loss": 0.5583,
+      "step": 3853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5744942254202638e-05,
+      "loss": 0.9416,
+      "step": 3854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574269527439581e-05,
+      "loss": 0.9149,
+      "step": 3855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5740447861891946e-05,
+      "loss": 0.9652,
+      "step": 3856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5738200016860382e-05,
+      "loss": 0.935,
+      "step": 3857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5735951739470485e-05,
+      "loss": 0.6292,
+      "step": 3858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573370302989166e-05,
+      "loss": 0.9232,
+      "step": 3859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731453888293343e-05,
+      "loss": 0.946,
+      "step": 3860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5729204314845002e-05,
+      "loss": 0.8997,
+      "step": 3861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5726954309716128e-05,
+      "loss": 0.9795,
+      "step": 3862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572470387307626e-05,
+      "loss": 0.953,
+      "step": 3863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572245300509496e-05,
+      "loss": 0.9352,
+      "step": 3864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720201705941827e-05,
+      "loss": 0.9493,
+      "step": 3865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571794997578649e-05,
+      "loss": 0.9096,
+      "step": 3866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5715697814798608e-05,
+      "loss": 0.9441,
+      "step": 3867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5713445223147876e-05,
+      "loss": 0.5029,
+      "step": 3868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5711192201004023e-05,
+      "loss": 0.9584,
+      "step": 3869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570893874853681e-05,
+      "loss": 0.9866,
+      "step": 3870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5706684865916025e-05,
+      "loss": 0.9917,
+      "step": 3871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5704430553311496e-05,
+      "loss": 0.9242,
+      "step": 3872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702175810893075e-05,
+      "loss": 0.9441,
+      "step": 3873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5699920638830656e-05,
+      "loss": 0.8917,
+      "step": 3874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5697665037294157e-05,
+      "loss": 0.921,
+      "step": 3875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5695409006453527e-05,
+      "loss": 0.9227,
+      "step": 3876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5693152546478762e-05,
+      "loss": 0.9315,
+      "step": 3877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569089565753987e-05,
+      "loss": 0.9906,
+      "step": 3878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5688638339806916e-05,
+      "loss": 0.8831,
+      "step": 3879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686380593449966e-05,
+      "loss": 0.9144,
+      "step": 3880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568412241863915e-05,
+      "loss": 0.9183,
+      "step": 3881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5681863815544604e-05,
+      "loss": 0.5891,
+      "step": 3882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5679604784336516e-05,
+      "loss": 0.9262,
+      "step": 3883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5677345325185093e-05,
+      "loss": 0.9787,
+      "step": 3884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675085438260584e-05,
+      "loss": 0.885,
+      "step": 3885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5672825123733257e-05,
+      "loss": 0.9154,
+      "step": 3886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567056438177343e-05,
+      "loss": 0.953,
+      "step": 3887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5668303212551438e-05,
+      "loss": 0.9373,
+      "step": 3888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566604161623766e-05,
+      "loss": 0.9551,
+      "step": 3889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5663779593002498e-05,
+      "loss": 0.8794,
+      "step": 3890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5661517143016386e-05,
+      "loss": 0.9182,
+      "step": 3891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56592542664498e-05,
+      "loss": 0.8903,
+      "step": 3892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565699096347324e-05,
+      "loss": 0.9366,
+      "step": 3893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565699096347324e-05,
+      "loss": 0.9324,
+      "step": 3894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5654727234257236e-05,
+      "loss": 0.9523,
+      "step": 3895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565246307897236e-05,
+      "loss": 0.8747,
+      "step": 3896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56501984977892e-05,
+      "loss": 0.9679,
+      "step": 3897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5647933490878402e-05,
+      "loss": 0.9637,
+      "step": 3898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5645668058410617e-05,
+      "loss": 0.9559,
+      "step": 3899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5643402200556543e-05,
+      "loss": 0.8751,
+      "step": 3900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5641135917486906e-05,
+      "loss": 0.9145,
+      "step": 3901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563886920937246e-05,
+      "loss": 0.945,
+      "step": 3902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5636602076384006e-05,
+      "loss": 0.9688,
+      "step": 3903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5634334518692355e-05,
+      "loss": 0.9535,
+      "step": 3904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5632066536468367e-05,
+      "loss": 0.9405,
+      "step": 3905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5629798129882926e-05,
+      "loss": 0.9751,
+      "step": 3906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562752929910696e-05,
+      "loss": 0.9443,
+      "step": 3907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5625260044311405e-05,
+      "loss": 0.9123,
+      "step": 3908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5622990365667256e-05,
+      "loss": 0.9698,
+      "step": 3909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5620720263345517e-05,
+      "loss": 0.9207,
+      "step": 3910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5618449737517242e-05,
+      "loss": 0.9271,
+      "step": 3911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5616178788353508e-05,
+      "loss": 0.8828,
+      "step": 3912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561390741602542e-05,
+      "loss": 0.9689,
+      "step": 3913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5611635620704128e-05,
+      "loss": 0.9114,
+      "step": 3914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5609363402560796e-05,
+      "loss": 0.9404,
+      "step": 3915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5607090761766634e-05,
+      "loss": 0.9384,
+      "step": 3916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604817698492886e-05,
+      "loss": 0.9471,
+      "step": 3917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5602544212910814e-05,
+      "loss": 0.9352,
+      "step": 3918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5600270305191715e-05,
+      "loss": 0.9363,
+      "step": 3919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5597995975506936e-05,
+      "loss": 0.9804,
+      "step": 3920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559572122402783e-05,
+      "loss": 0.9198,
+      "step": 3921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593446050925797e-05,
+      "loss": 0.9632,
+      "step": 3922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5591170456372264e-05,
+      "loss": 0.9684,
+      "step": 3923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5588894440538694e-05,
+      "loss": 0.9154,
+      "step": 3924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558661800359658e-05,
+      "loss": 0.9294,
+      "step": 3925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558434114571744e-05,
+      "loss": 0.9449,
+      "step": 3926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5582063867072835e-05,
+      "loss": 0.9902,
+      "step": 3927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5579786167834345e-05,
+      "loss": 0.9385,
+      "step": 3928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5577508048173596e-05,
+      "loss": 0.9502,
+      "step": 3929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5575229508262237e-05,
+      "loss": 0.9481,
+      "step": 3930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5572950548271946e-05,
+      "loss": 0.9557,
+      "step": 3931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557067116837444e-05,
+      "loss": 0.8999,
+      "step": 3932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556839136874146e-05,
+      "loss": 0.965,
+      "step": 3933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5566111149544792e-05,
+      "loss": 0.924,
+      "step": 3934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5563830510956234e-05,
+      "loss": 0.8841,
+      "step": 3935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556154945314763e-05,
+      "loss": 0.944,
+      "step": 3936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5559267976290854e-05,
+      "loss": 0.9637,
+      "step": 3937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555698608055781e-05,
+      "loss": 0.9092,
+      "step": 3938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555470376612043e-05,
+      "loss": 0.9483,
+      "step": 3939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555242103315068e-05,
+      "loss": 0.7475,
+      "step": 3940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555013788182056e-05,
+      "loss": 0.9506,
+      "step": 3941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5547854312302096e-05,
+      "loss": 0.9548,
+      "step": 3942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5545570324767357e-05,
+      "loss": 0.9369,
+      "step": 3943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5543285919388426e-05,
+      "loss": 0.9484,
+      "step": 3944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5541001096337433e-05,
+      "loss": 0.9589,
+      "step": 3945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538715855786526e-05,
+      "loss": 0.9541,
+      "step": 3946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5536430197907904e-05,
+      "loss": 1.008,
+      "step": 3947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553414412287378e-05,
+      "loss": 0.9517,
+      "step": 3948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5531857630856395e-05,
+      "loss": 0.9257,
+      "step": 3949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552957072202804e-05,
+      "loss": 0.9289,
+      "step": 3950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527283396561025e-05,
+      "loss": 0.9377,
+      "step": 3951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5524995654627696e-05,
+      "loss": 0.9925,
+      "step": 3952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5522707496400425e-05,
+      "loss": 0.8908,
+      "step": 3953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5520418922051624e-05,
+      "loss": 1.0111,
+      "step": 3954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5518129931753724e-05,
+      "loss": 0.9178,
+      "step": 3955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.55158405256792e-05,
+      "loss": 0.9766,
+      "step": 3956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5513550704000548e-05,
+      "loss": 0.9778,
+      "step": 3957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5511260466890302e-05,
+      "loss": 0.9468,
+      "step": 3958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5508969814521026e-05,
+      "loss": 0.9725,
+      "step": 3959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5506678747065314e-05,
+      "loss": 0.9484,
+      "step": 3960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5504387264695795e-05,
+      "loss": 0.9577,
+      "step": 3961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502095367585124e-05,
+      "loss": 0.9232,
+      "step": 3962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549980305590599e-05,
+      "loss": 0.9858,
+      "step": 3963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5497510329831105e-05,
+      "loss": 0.9318,
+      "step": 3964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549521718953323e-05,
+      "loss": 0.9328,
+      "step": 3965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549292363518515e-05,
+      "loss": 0.936,
+      "step": 3966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5490629666959668e-05,
+      "loss": 1.0057,
+      "step": 3967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548833528502963e-05,
+      "loss": 0.9355,
+      "step": 3968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5486040489567916e-05,
+      "loss": 0.944,
+      "step": 3969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5483745280747432e-05,
+      "loss": 0.9576,
+      "step": 3970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5481449658741112e-05,
+      "loss": 0.9186,
+      "step": 3971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547915362372193e-05,
+      "loss": 0.9356,
+      "step": 3972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547685717586288e-05,
+      "loss": 0.9501,
+      "step": 3973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5474560315337007e-05,
+      "loss": 0.9529,
+      "step": 3974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472263042317356e-05,
+      "loss": 0.949,
+      "step": 3975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469965356977035e-05,
+      "loss": 0.9041,
+      "step": 3976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5467667259489157e-05,
+      "loss": 0.8782,
+      "step": 3977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5465368750026887e-05,
+      "loss": 0.9516,
+      "step": 3978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463069828763405e-05,
+      "loss": 0.9055,
+      "step": 3979
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546077049587193e-05,
+      "loss": 0.9133,
+      "step": 3980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5458470751525716e-05,
+      "loss": 0.9107,
+      "step": 3981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456170595898033e-05,
+      "loss": 0.9646,
+      "step": 3982
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5453870029162202e-05,
+      "loss": 0.9651,
+      "step": 3983
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5451569051491558e-05,
+      "loss": 0.8791,
+      "step": 3984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5449267663059476e-05,
+      "loss": 0.9205,
+      "step": 3985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5446965864039357e-05,
+      "loss": 0.9494,
+      "step": 3986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5444663654604643e-05,
+      "loss": 0.9272,
+      "step": 3987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544236103492879e-05,
+      "loss": 0.8699,
+      "step": 3988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5440058005185295e-05,
+      "loss": 0.938,
+      "step": 3989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5437754565547694e-05,
+      "loss": 0.9033,
+      "step": 3990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543545071618954e-05,
+      "loss": 0.9404,
+      "step": 3991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543314645728442e-05,
+      "loss": 0.9262,
+      "step": 3992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430841789005956e-05,
+      "loss": 0.9512,
+      "step": 3993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54285367115278e-05,
+      "loss": 0.9461,
+      "step": 3994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542623122502363e-05,
+      "loss": 0.9559,
+      "step": 3995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5423925329667163e-05,
+      "loss": 0.9014,
+      "step": 3996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542161902563214e-05,
+      "loss": 0.938,
+      "step": 3997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5419312313092328e-05,
+      "loss": 0.9903,
+      "step": 3998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5417005192221542e-05,
+      "loss": 0.9568,
+      "step": 3999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5414697663193616e-05,
+      "loss": 0.9419,
+      "step": 4000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541238972618241e-05,
+      "loss": 0.951,
+      "step": 4001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541008138136183e-05,
+      "loss": 0.9507,
+      "step": 4002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5407772628905794e-05,
+      "loss": 0.952,
+      "step": 4003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540546346898827e-05,
+      "loss": 0.9503,
+      "step": 4004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403153901783243e-05,
+      "loss": 0.9487,
+      "step": 4005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5400843927464727e-05,
+      "loss": 0.9466,
+      "step": 4006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539853354620678e-05,
+      "loss": 0.9386,
+      "step": 4007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396222758183477e-05,
+      "loss": 0.9266,
+      "step": 4008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393911563568938e-05,
+      "loss": 0.9179,
+      "step": 4009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53915999625373e-05,
+      "loss": 0.9222,
+      "step": 4010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538928795526274e-05,
+      "loss": 0.9168,
+      "step": 4011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386975541919457e-05,
+      "loss": 0.9462,
+      "step": 4012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5384662722681688e-05,
+      "loss": 0.9271,
+      "step": 4013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5382349497723693e-05,
+      "loss": 1.0084,
+      "step": 4014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5380035867219776e-05,
+      "loss": 0.9656,
+      "step": 4015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377721831344258e-05,
+      "loss": 1.0007,
+      "step": 4016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5375407390271492e-05,
+      "loss": 1.0028,
+      "step": 4017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5373092544175875e-05,
+      "loss": 0.9863,
+      "step": 4018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370777293231814e-05,
+      "loss": 0.9377,
+      "step": 4019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5368461637613762e-05,
+      "loss": 0.9391,
+      "step": 4020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53661455774962e-05,
+      "loss": 0.9802,
+      "step": 4021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5363829113053633e-05,
+      "loss": 0.9139,
+      "step": 4022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5361512244460605e-05,
+      "loss": 0.961,
+      "step": 4023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535919497189168e-05,
+      "loss": 0.9191,
+      "step": 4024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535687729552146e-05,
+      "loss": 0.9677,
+      "step": 4025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5354559215524577e-05,
+      "loss": 0.8915,
+      "step": 4026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5352240732075694e-05,
+      "loss": 0.8928,
+      "step": 4027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349921845349504e-05,
+      "loss": 0.9413,
+      "step": 4028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5347602555520725e-05,
+      "loss": 0.9443,
+      "step": 4029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534528286276411e-05,
+      "loss": 0.9264,
+      "step": 4030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534296276725444e-05,
+      "loss": 0.9318,
+      "step": 4031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534064226916653e-05,
+      "loss": 0.9494,
+      "step": 4032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5338321368675226e-05,
+      "loss": 0.9542,
+      "step": 4033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53360000659554e-05,
+      "loss": 0.927,
+      "step": 4034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5333678361181955e-05,
+      "loss": 0.9225,
+      "step": 4035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5331356254529826e-05,
+      "loss": 0.9002,
+      "step": 4036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5329033746173975e-05,
+      "loss": 0.9351,
+      "step": 4037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5326710836289406e-05,
+      "loss": 0.9156,
+      "step": 4038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532438752505113e-05,
+      "loss": 0.9352,
+      "step": 4039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322063812634213e-05,
+      "loss": 0.9734,
+      "step": 4040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5319739699213743e-05,
+      "loss": 0.985,
+      "step": 4041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5317415184964824e-05,
+      "loss": 0.9665,
+      "step": 4042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315090270062612e-05,
+      "loss": 0.9765,
+      "step": 4043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531276495468228e-05,
+      "loss": 0.9196,
+      "step": 4044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531043923899903e-05,
+      "loss": 0.9448,
+      "step": 4045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530811312318811e-05,
+      "loss": 0.8861,
+      "step": 4046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530578660742478e-05,
+      "loss": 0.9204,
+      "step": 4047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5303459691884333e-05,
+      "loss": 0.8889,
+      "step": 4048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5301132376742097e-05,
+      "loss": 0.9139,
+      "step": 4049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529880466217344e-05,
+      "loss": 0.9675,
+      "step": 4050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5296476548353734e-05,
+      "loss": 0.9444,
+      "step": 4051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294148035458406e-05,
+      "loss": 0.9424,
+      "step": 4052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5291819123662897e-05,
+      "loss": 0.9562,
+      "step": 4053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528948981314269e-05,
+      "loss": 0.9478,
+      "step": 4054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528716010407329e-05,
+      "loss": 0.9249,
+      "step": 4055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284829996630234e-05,
+      "loss": 0.9664,
+      "step": 4056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282499490989094e-05,
+      "loss": 0.9512,
+      "step": 4057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5280168587325462e-05,
+      "loss": 0.9905,
+      "step": 4058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5277837285814964e-05,
+      "loss": 0.9205,
+      "step": 4059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5275505586633264e-05,
+      "loss": 0.8971,
+      "step": 4060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273173489956045e-05,
+      "loss": 0.541,
+      "step": 4061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5270840995959028e-05,
+      "loss": 0.9374,
+      "step": 4062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5268508104817955e-05,
+      "loss": 0.8754,
+      "step": 4063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5266174816708607e-05,
+      "loss": 0.9612,
+      "step": 4064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526384113180679e-05,
+      "loss": 0.9492,
+      "step": 4065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526150705028834e-05,
+      "loss": 0.9473,
+      "step": 4066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5259172572329132e-05,
+      "loss": 0.9929,
+      "step": 4067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5256837698105047e-05,
+      "loss": 0.9827,
+      "step": 4068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5254502427792025e-05,
+      "loss": 1.0001,
+      "step": 4069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252166761566018e-05,
+      "loss": 0.9373,
+      "step": 4070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5249830699603014e-05,
+      "loss": 0.9577,
+      "step": 4071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5247494242079024e-05,
+      "loss": 0.96,
+      "step": 4072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5245157389170099e-05,
+      "loss": 0.9462,
+      "step": 4073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5242820141052313e-05,
+      "loss": 0.8889,
+      "step": 4074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5240482497901772e-05,
+      "loss": 0.5374,
+      "step": 4075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5238144459894612e-05,
+      "loss": 0.9544,
+      "step": 4076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5235806027206997e-05,
+      "loss": 0.922,
+      "step": 4077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.523346720001512e-05,
+      "loss": 0.917,
+      "step": 4078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5231127978495208e-05,
+      "loss": 0.9106,
+      "step": 4079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5228788362823515e-05,
+      "loss": 0.93,
+      "step": 4080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226448353176327e-05,
+      "loss": 0.5844,
+      "step": 4081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5224107949729952e-05,
+      "loss": 0.9394,
+      "step": 4082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221767152660736e-05,
+      "loss": 0.9623,
+      "step": 4083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5219425962145054e-05,
+      "loss": 0.9212,
+      "step": 4084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217084378359306e-05,
+      "loss": 0.9794,
+      "step": 4085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214742401479926e-05,
+      "loss": 0.9091,
+      "step": 4086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5212400031683374e-05,
+      "loss": 0.909,
+      "step": 4087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5210057269146141e-05,
+      "loss": 0.5697,
+      "step": 4088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207714114044752e-05,
+      "loss": 0.9426,
+      "step": 4089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5205370566555756e-05,
+      "loss": 0.9159,
+      "step": 4090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5203026626855728e-05,
+      "loss": 0.9792,
+      "step": 4091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5200682295121286e-05,
+      "loss": 0.9188,
+      "step": 4092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519833757152906e-05,
+      "loss": 0.9032,
+      "step": 4093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195992456255728e-05,
+      "loss": 0.8805,
+      "step": 4094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5193646949477983e-05,
+      "loss": 0.8987,
+      "step": 4095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5191301051372554e-05,
+      "loss": 0.9747,
+      "step": 4096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188954762116197e-05,
+      "loss": 0.9029,
+      "step": 4097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.51866080818857e-05,
+      "loss": 0.9461,
+      "step": 4098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5184261010857881e-05,
+      "loss": 0.9128,
+      "step": 4099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5181913549209582e-05,
+      "loss": 0.903,
+      "step": 4100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5179565697117679e-05,
+      "loss": 0.9259,
+      "step": 4101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5177217454759081e-05,
+      "loss": 0.9164,
+      "step": 4102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5174868822310715e-05,
+      "loss": 0.9996,
+      "step": 4103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517251979994955e-05,
+      "loss": 0.9541,
+      "step": 4104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170170387852574e-05,
+      "loss": 0.9108,
+      "step": 4105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516782058619681e-05,
+      "loss": 0.8808,
+      "step": 4106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5165470395159314e-05,
+      "loss": 0.9112,
+      "step": 4107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163119814917162e-05,
+      "loss": 0.91,
+      "step": 4108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5160768845647464e-05,
+      "loss": 0.9218,
+      "step": 4109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158417487527364e-05,
+      "loss": 0.9,
+      "step": 4110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5156065740734027e-05,
+      "loss": 0.9385,
+      "step": 4111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515371360544465e-05,
+      "loss": 0.914,
+      "step": 4112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5151361081836461e-05,
+      "loss": 0.9259,
+      "step": 4113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5149008170086722e-05,
+      "loss": 0.9763,
+      "step": 4114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514665487037271e-05,
+      "loss": 0.9434,
+      "step": 4115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5144301182871743e-05,
+      "loss": 0.9271,
+      "step": 4116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5141947107761166e-05,
+      "loss": 0.9421,
+      "step": 4117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5139592645218355e-05,
+      "loss": 0.992,
+      "step": 4118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5137237795420707e-05,
+      "loss": 0.9397,
+      "step": 4119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513488255854566e-05,
+      "loss": 0.91,
+      "step": 4120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513252693477067e-05,
+      "loss": 0.9389,
+      "step": 4121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5130170924273229e-05,
+      "loss": 0.9287,
+      "step": 4122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5127814527230858e-05,
+      "loss": 0.9161,
+      "step": 4123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5125457743821098e-05,
+      "loss": 0.9517,
+      "step": 4124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123100574221537e-05,
+      "loss": 1.0113,
+      "step": 4125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120743018609773e-05,
+      "loss": 0.9282,
+      "step": 4126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118385077163446e-05,
+      "loss": 0.9353,
+      "step": 4127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5116026750060223e-05,
+      "loss": 0.9115,
+      "step": 4128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5113668037477789e-05,
+      "loss": 0.9207,
+      "step": 4129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5111308939593876e-05,
+      "loss": 0.9102,
+      "step": 4130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510894945658623e-05,
+      "loss": 0.9093,
+      "step": 4131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5106589588632636e-05,
+      "loss": 0.9221,
+      "step": 4132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104229335910901e-05,
+      "loss": 0.9061,
+      "step": 4133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101868698598867e-05,
+      "loss": 0.949,
+      "step": 4134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099507676874397e-05,
+      "loss": 0.8787,
+      "step": 4135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5097146270915391e-05,
+      "loss": 0.9177,
+      "step": 4136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094784480899774e-05,
+      "loss": 0.9416,
+      "step": 4137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50924223070055e-05,
+      "loss": 0.9902,
+      "step": 4138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5090059749410553e-05,
+      "loss": 0.8943,
+      "step": 4139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5087696808292947e-05,
+      "loss": 0.9255,
+      "step": 4140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085333483830721e-05,
+      "loss": 0.8981,
+      "step": 4141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082969776201948e-05,
+      "loss": 0.8801,
+      "step": 4142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080605685584723e-05,
+      "loss": 0.9585,
+      "step": 4143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507824121215718e-05,
+      "loss": 0.907,
+      "step": 4144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075876356097472e-05,
+      "loss": 0.8754,
+      "step": 4145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5073511117583786e-05,
+      "loss": 0.9463,
+      "step": 4146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5071145496794338e-05,
+      "loss": 0.5075,
+      "step": 4147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068779493907364e-05,
+      "loss": 0.9291,
+      "step": 4148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066413109101143e-05,
+      "loss": 0.8927,
+      "step": 4149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064046342553976e-05,
+      "loss": 0.9361,
+      "step": 4150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506167919444419e-05,
+      "loss": 0.9598,
+      "step": 4151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059311664950146e-05,
+      "loss": 0.9617,
+      "step": 4152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5056943754250226e-05,
+      "loss": 0.9704,
+      "step": 4153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505457546252285e-05,
+      "loss": 0.8867,
+      "step": 4154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052206789946467e-05,
+      "loss": 0.9282,
+      "step": 4155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5049837736699542e-05,
+      "loss": 0.9,
+      "step": 4156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047468302960577e-05,
+      "loss": 0.8946,
+      "step": 4157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045098488908107e-05,
+      "loss": 0.8893,
+      "step": 4158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5042728294720692e-05,
+      "loss": 0.9806,
+      "step": 4159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040357720576917e-05,
+      "loss": 0.9486,
+      "step": 4160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50379867666554e-05,
+      "loss": 0.9411,
+      "step": 4161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5035615433134787e-05,
+      "loss": 0.9415,
+      "step": 4162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5033243720193746e-05,
+      "loss": 0.971,
+      "step": 4163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5030871628010987e-05,
+      "loss": 0.9821,
+      "step": 4164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502849915676524e-05,
+      "loss": 0.9203,
+      "step": 4165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5026126306635256e-05,
+      "loss": 0.9392,
+      "step": 4166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023753077799834e-05,
+      "loss": 0.5909,
+      "step": 4167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021379470437785e-05,
+      "loss": 0.9491,
+      "step": 4168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019005484727953e-05,
+      "loss": 0.5564,
+      "step": 4169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5016631120849217e-05,
+      "loss": 0.9153,
+      "step": 4170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5014256378980471e-05,
+      "loss": 0.9534,
+      "step": 4171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5011881259300654e-05,
+      "loss": 0.9178,
+      "step": 4172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5009505761988718e-05,
+      "loss": 0.9272,
+      "step": 4173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5007129887223654e-05,
+      "loss": 0.9377,
+      "step": 4174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5004753635184482e-05,
+      "loss": 0.9323,
+      "step": 4175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002377006050242e-05,
+      "loss": 0.9467,
+      "step": 4176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.9711,
+      "step": 4177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499762261721287e-05,
+      "loss": 0.9292,
+      "step": 4178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4995244857867973e-05,
+      "loss": 0.8948,
+      "step": 4179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.499286672214447e-05,
+      "loss": 0.9759,
+      "step": 4180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4990488210221545e-05,
+      "loss": 0.899,
+      "step": 4181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4988109322278417e-05,
+      "loss": 0.9137,
+      "step": 4182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4985730058494325e-05,
+      "loss": 0.9678,
+      "step": 4183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4983350419048544e-05,
+      "loss": 0.9176,
+      "step": 4184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980970404120367e-05,
+      "loss": 0.9774,
+      "step": 4185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497859001388913e-05,
+      "loss": 0.9414,
+      "step": 4186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4976209248534183e-05,
+      "loss": 0.93,
+      "step": 4187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4973828108234914e-05,
+      "loss": 0.9582,
+      "step": 4188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4971446593170733e-05,
+      "loss": 0.9292,
+      "step": 4189
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4969064703521082e-05,
+      "loss": 0.886,
+      "step": 4190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4966682439465432e-05,
+      "loss": 0.5775,
+      "step": 4191
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4964299801183278e-05,
+      "loss": 0.8678,
+      "step": 4192
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4961916788854147e-05,
+      "loss": 0.9243,
+      "step": 4193
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4959533402657591e-05,
+      "loss": 0.9456,
+      "step": 4194
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4957149642773195e-05,
+      "loss": 0.9289,
+      "step": 4195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4954765509380565e-05,
+      "loss": 0.9238,
+      "step": 4196
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4952381002659344e-05,
+      "loss": 0.9489,
+      "step": 4197
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4949996122789193e-05,
+      "loss": 0.9226,
+      "step": 4198
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947610869949811e-05,
+      "loss": 0.9226,
+      "step": 4199
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945225244320916e-05,
+      "loss": 0.9646,
+      "step": 4200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4942839246082264e-05,
+      "loss": 0.9485,
+      "step": 4201
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940452875413627e-05,
+      "loss": 0.9528,
+      "step": 4202
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493806613249482e-05,
+      "loss": 0.9122,
+      "step": 4203
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935679017505671e-05,
+      "loss": 0.9358,
+      "step": 4204
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933291530626047e-05,
+      "loss": 0.9527,
+      "step": 4205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930903672035835e-05,
+      "loss": 0.9061,
+      "step": 4206
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492851544191496e-05,
+      "loss": 0.9322,
+      "step": 4207
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492612684044336e-05,
+      "loss": 0.9434,
+      "step": 4208
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4923737867801018e-05,
+      "loss": 0.8712,
+      "step": 4209
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4921348524167934e-05,
+      "loss": 0.961,
+      "step": 4210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918958809724135e-05,
+      "loss": 0.964,
+      "step": 4211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916568724649688e-05,
+      "loss": 0.9342,
+      "step": 4212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491417826912467e-05,
+      "loss": 0.5095,
+      "step": 4213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4911787443329204e-05,
+      "loss": 0.9313,
+      "step": 4214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4909396247443429e-05,
+      "loss": 0.9407,
+      "step": 4215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907004681647515e-05,
+      "loss": 0.9155,
+      "step": 4216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904612746121657e-05,
+      "loss": 0.9518,
+      "step": 4217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4902220441046089e-05,
+      "loss": 0.5944,
+      "step": 4218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489982776660106e-05,
+      "loss": 0.9002,
+      "step": 4219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4897434722966851e-05,
+      "loss": 0.8893,
+      "step": 4220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4895041310323772e-05,
+      "loss": 0.971,
+      "step": 4221
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489264752885216e-05,
+      "loss": 0.9285,
+      "step": 4222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890253378732385e-05,
+      "loss": 0.9388,
+      "step": 4223
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887858860144833e-05,
+      "loss": 0.9773,
+      "step": 4224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4885463973269929e-05,
+      "loss": 0.9234,
+      "step": 4225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4883068718288121e-05,
+      "loss": 0.9199,
+      "step": 4226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4880673095379883e-05,
+      "loss": 0.9751,
+      "step": 4227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878277104725722e-05,
+      "loss": 0.931,
+      "step": 4228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875880746506169e-05,
+      "loss": 0.9693,
+      "step": 4229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4873484020901778e-05,
+      "loss": 0.9156,
+      "step": 4230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4871086928093142e-05,
+      "loss": 0.9337,
+      "step": 4231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4868689468260876e-05,
+      "loss": 0.9963,
+      "step": 4232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486629164158562e-05,
+      "loss": 0.9301,
+      "step": 4233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4863893448248043e-05,
+      "loss": 0.9141,
+      "step": 4234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861494888428845e-05,
+      "loss": 0.9571,
+      "step": 4235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4859095962308749e-05,
+      "loss": 0.9171,
+      "step": 4236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856696670068508e-05,
+      "loss": 0.8651,
+      "step": 4237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4854297011888905e-05,
+      "loss": 0.9043,
+      "step": 4238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851896987950749e-05,
+      "loss": 0.8992,
+      "step": 4239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4849496598434869e-05,
+      "loss": 0.9131,
+      "step": 4240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847095843522133e-05,
+      "loss": 0.9883,
+      "step": 4241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4844694723393429e-05,
+      "loss": 0.9675,
+      "step": 4242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842293238229678e-05,
+      "loss": 0.9533,
+      "step": 4243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4839891388211822e-05,
+      "loss": 0.909,
+      "step": 4244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837489173520838e-05,
+      "loss": 0.9146,
+      "step": 4245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4835086594337725e-05,
+      "loss": 0.9135,
+      "step": 4246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.483268365084351e-05,
+      "loss": 0.9595,
+      "step": 4247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4830280343219247e-05,
+      "loss": 0.9274,
+      "step": 4248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4827876671646026e-05,
+      "loss": 0.946,
+      "step": 4249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4825472636304952e-05,
+      "loss": 0.9126,
+      "step": 4250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823068237377161e-05,
+      "loss": 0.9366,
+      "step": 4251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4820663475043818e-05,
+      "loss": 0.9222,
+      "step": 4252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818258349486121e-05,
+      "loss": 0.9216,
+      "step": 4253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815852860885287e-05,
+      "loss": 0.9011,
+      "step": 4254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4813447009422564e-05,
+      "loss": 0.9199,
+      "step": 4255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811040795279223e-05,
+      "loss": 0.956,
+      "step": 4256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480863421863657e-05,
+      "loss": 0.9356,
+      "step": 4257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4806227279675933e-05,
+      "loss": 0.946,
+      "step": 4258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480381997857867e-05,
+      "loss": 0.947,
+      "step": 4259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4801412315526164e-05,
+      "loss": 0.5727,
+      "step": 4260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799004290699825e-05,
+      "loss": 0.9373,
+      "step": 4261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479659590428109e-05,
+      "loss": 0.6513,
+      "step": 4262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4794187156451428e-05,
+      "loss": 0.9278,
+      "step": 4263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479177804739233e-05,
+      "loss": 0.9435,
+      "step": 4264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4789368577285314e-05,
+      "loss": 0.9407,
+      "step": 4265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4786958746311932e-05,
+      "loss": 0.9433,
+      "step": 4266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4784548554653756e-05,
+      "loss": 1.0098,
+      "step": 4267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782138002492385e-05,
+      "loss": 0.9394,
+      "step": 4268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4779727090009452e-05,
+      "loss": 0.9027,
+      "step": 4269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777315817386616e-05,
+      "loss": 0.9554,
+      "step": 4270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.477490418480555e-05,
+      "loss": 0.9669,
+      "step": 4271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4772492192447969e-05,
+      "loss": 0.6318,
+      "step": 4272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4770079840495615e-05,
+      "loss": 0.9152,
+      "step": 4273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4767667129130243e-05,
+      "loss": 0.9288,
+      "step": 4274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4765254058533651e-05,
+      "loss": 0.9292,
+      "step": 4275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4762840628887658e-05,
+      "loss": 0.9352,
+      "step": 4276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476042684037411e-05,
+      "loss": 0.9553,
+      "step": 4277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4758012693174875e-05,
+      "loss": 0.9858,
+      "step": 4278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755598187471853e-05,
+      "loss": 0.9436,
+      "step": 4279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4753183323446978e-05,
+      "loss": 0.9223,
+      "step": 4280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4750768101282198e-05,
+      "loss": 0.9383,
+      "step": 4281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4748352521159492e-05,
+      "loss": 0.9522,
+      "step": 4282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474593658326087e-05,
+      "loss": 0.9053,
+      "step": 4283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4743520287768366e-05,
+      "loss": 0.9468,
+      "step": 4284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741103634864042e-05,
+      "loss": 0.9635,
+      "step": 4285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738686624729987e-05,
+      "loss": 0.8989,
+      "step": 4286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736269257548316e-05,
+      "loss": 0.9536,
+      "step": 4287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4733851533501172e-05,
+      "loss": 0.9221,
+      "step": 4288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4731433452770723e-05,
+      "loss": 0.9618,
+      "step": 4289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4729015015539165e-05,
+      "loss": 0.976,
+      "step": 4290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4726596221988722e-05,
+      "loss": 0.9323,
+      "step": 4291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4724177072301642e-05,
+      "loss": 0.9168,
+      "step": 4292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4721757566660203e-05,
+      "loss": 0.906,
+      "step": 4293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719337705246707e-05,
+      "loss": 0.946,
+      "step": 4294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716917488243485e-05,
+      "loss": 0.9358,
+      "step": 4295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4714496915832892e-05,
+      "loss": 0.9351,
+      "step": 4296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712075988197318e-05,
+      "loss": 0.8978,
+      "step": 4297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709654705519168e-05,
+      "loss": 0.935,
+      "step": 4298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4707233067980881e-05,
+      "loss": 0.9478,
+      "step": 4299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4704811075764922e-05,
+      "loss": 0.9093,
+      "step": 4300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470238872905378e-05,
+      "loss": 0.9615,
+      "step": 4301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4699966028029976e-05,
+      "loss": 0.9262,
+      "step": 4302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697542972876047e-05,
+      "loss": 0.9732,
+      "step": 4303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4695119563774568e-05,
+      "loss": 0.884,
+      "step": 4304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469269580090814e-05,
+      "loss": 0.9475,
+      "step": 4305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469027168445938e-05,
+      "loss": 0.8994,
+      "step": 4306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4687847214610944e-05,
+      "loss": 0.8973,
+      "step": 4307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.468542239154551e-05,
+      "loss": 0.9806,
+      "step": 4308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.468299721544578e-05,
+      "loss": 0.9412,
+      "step": 4309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4680571686494483e-05,
+      "loss": 0.977,
+      "step": 4310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4678145804874377e-05,
+      "loss": 0.9596,
+      "step": 4311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4675719570768252e-05,
+      "loss": 0.9085,
+      "step": 4312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4673292984358911e-05,
+      "loss": 0.9564,
+      "step": 4313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4670866045829191e-05,
+      "loss": 0.9232,
+      "step": 4314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466843875536196e-05,
+      "loss": 0.9692,
+      "step": 4315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666011113140103e-05,
+      "loss": 0.9572,
+      "step": 4316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466358311934654e-05,
+      "loss": 0.934,
+      "step": 4317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661154774164213e-05,
+      "loss": 0.9186,
+      "step": 4318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4658726077776093e-05,
+      "loss": 0.9485,
+      "step": 4319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656297030365175e-05,
+      "loss": 1.0001,
+      "step": 4320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653867632114478e-05,
+      "loss": 0.9417,
+      "step": 4321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651437883207056e-05,
+      "loss": 0.9672,
+      "step": 4322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464900778382598e-05,
+      "loss": 0.9531,
+      "step": 4323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646577334154355e-05,
+      "loss": 0.923,
+      "step": 4324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4644146534375307e-05,
+      "loss": 0.8964,
+      "step": 4325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464171538467199e-05,
+      "loss": 0.9405,
+      "step": 4326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4639283885227586e-05,
+      "loss": 0.9448,
+      "step": 4327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636852036225304e-05,
+      "loss": 0.9599,
+      "step": 4328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4634419837848373e-05,
+      "loss": 0.9322,
+      "step": 4329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4631987290280054e-05,
+      "loss": 0.9171,
+      "step": 4330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629554393703635e-05,
+      "loss": 0.9231,
+      "step": 4331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627121148302428e-05,
+      "loss": 0.9178,
+      "step": 4332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624687554259773e-05,
+      "loss": 0.9214,
+      "step": 4333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462225361175903e-05,
+      "loss": 0.9306,
+      "step": 4334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4619819320983593e-05,
+      "loss": 0.934,
+      "step": 4335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617384682116882e-05,
+      "loss": 0.923,
+      "step": 4336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4614949695342335e-05,
+      "loss": 0.9318,
+      "step": 4337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612514360843426e-05,
+      "loss": 0.8914,
+      "step": 4338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4610078678803651e-05,
+      "loss": 0.9347,
+      "step": 4339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4607642649406529e-05,
+      "loss": 0.9119,
+      "step": 4340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4605206272835614e-05,
+      "loss": 0.9214,
+      "step": 4341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602769549274477e-05,
+      "loss": 0.9424,
+      "step": 4342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600332478906718e-05,
+      "loss": 0.9288,
+      "step": 4343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597895061915968e-05,
+      "loss": 0.9289,
+      "step": 4344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595457298485873e-05,
+      "loss": 0.9431,
+      "step": 4345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459301918880012e-05,
+      "loss": 0.9109,
+      "step": 4346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4590580733042408e-05,
+      "loss": 0.9291,
+      "step": 4347
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458814193139647e-05,
+      "loss": 0.9587,
+      "step": 4348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4585702784046065e-05,
+      "loss": 0.929,
+      "step": 4349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583263291174974e-05,
+      "loss": 0.905,
+      "step": 4350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.458082345296701e-05,
+      "loss": 0.5237,
+      "step": 4351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578383269606004e-05,
+      "loss": 0.9561,
+      "step": 4352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575942741275819e-05,
+      "loss": 0.5217,
+      "step": 4353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4573501868160347e-05,
+      "loss": 0.9303,
+      "step": 4354
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571060650443494e-05,
+      "loss": 0.8763,
+      "step": 4355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568619088309204e-05,
+      "loss": 0.9003,
+      "step": 4356
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566177181941439e-05,
+      "loss": 0.9083,
+      "step": 4357
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4563734931524191e-05,
+      "loss": 0.9349,
+      "step": 4358
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4561292337241481e-05,
+      "loss": 0.5032,
+      "step": 4359
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558849399277348e-05,
+      "loss": 0.9363,
+      "step": 4360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4556406117815864e-05,
+      "loss": 0.9548,
+      "step": 4361
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455396249304112e-05,
+      "loss": 0.9684,
+      "step": 4362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551518525137241e-05,
+      "loss": 0.5235,
+      "step": 4363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549074214288368e-05,
+      "loss": 0.9526,
+      "step": 4364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546629560678682e-05,
+      "loss": 0.9539,
+      "step": 4365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4544184564492373e-05,
+      "loss": 0.9089,
+      "step": 4366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4541739225913669e-05,
+      "loss": 0.9289,
+      "step": 4367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453929354512682e-05,
+      "loss": 0.6859,
+      "step": 4368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.45368475223161e-05,
+      "loss": 0.87,
+      "step": 4369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534401157665812e-05,
+      "loss": 0.9558,
+      "step": 4370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531954451360281e-05,
+      "loss": 0.9474,
+      "step": 4371
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4529507403583862e-05,
+      "loss": 0.9343,
+      "step": 4372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4527060014520932e-05,
+      "loss": 0.8781,
+      "step": 4373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524612284355897e-05,
+      "loss": 0.8955,
+      "step": 4374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522164213273187e-05,
+      "loss": 0.9357,
+      "step": 4375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519715801457256e-05,
+      "loss": 0.9237,
+      "step": 4376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517267049092585e-05,
+      "loss": 0.5617,
+      "step": 4377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4514817956363684e-05,
+      "loss": 0.8634,
+      "step": 4378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4512368523455085e-05,
+      "loss": 0.9343,
+      "step": 4379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4509918750551343e-05,
+      "loss": 0.9618,
+      "step": 4380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4507468637837048e-05,
+      "loss": 0.9532,
+      "step": 4381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505018185496802e-05,
+      "loss": 0.9809,
+      "step": 4382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4502567393715246e-05,
+      "loss": 0.937,
+      "step": 4383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500116262677037e-05,
+      "loss": 0.9614,
+      "step": 4384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4497664792566865e-05,
+      "loss": 0.9381,
+      "step": 4385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495212983569445e-05,
+      "loss": 0.9056,
+      "step": 4386
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4492760835869504e-05,
+      "loss": 0.9284,
+      "step": 4387
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490308349651812e-05,
+      "loss": 0.944,
+      "step": 4388
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487855525101156e-05,
+      "loss": 0.93,
+      "step": 4389
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4485402362402348e-05,
+      "loss": 0.9171,
+      "step": 4390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482948861740229e-05,
+      "loss": 0.893,
+      "step": 4391
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4480495023299665e-05,
+      "loss": 0.9648,
+      "step": 4392
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478040847265546e-05,
+      "loss": 0.9419,
+      "step": 4393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4475586333822787e-05,
+      "loss": 0.9309,
+      "step": 4394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4473131483156326e-05,
+      "loss": 0.9228,
+      "step": 4395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470676295451138e-05,
+      "loss": 1.0045,
+      "step": 4396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4468220770892208e-05,
+      "loss": 0.9348,
+      "step": 4397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4465764909664552e-05,
+      "loss": 0.9579,
+      "step": 4398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446330871195322e-05,
+      "loss": 0.9086,
+      "step": 4399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446085217794327e-05,
+      "loss": 1.014,
+      "step": 4400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4458395307819806e-05,
+      "loss": 0.9237,
+      "step": 4401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445593810176794e-05,
+      "loss": 0.8237,
+      "step": 4402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4453480559972817e-05,
+      "loss": 0.9296,
+      "step": 4403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4451022682619606e-05,
+      "loss": 0.8827,
+      "step": 4404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448564469893507e-05,
+      "loss": 0.9172,
+      "step": 4405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4446105921979731e-05,
+      "loss": 0.9625,
+      "step": 4406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4443647039063534e-05,
+      "loss": 0.9436,
+      "step": 4407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4441187821330176e-05,
+      "loss": 0.9199,
+      "step": 4408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438728268964956e-05,
+      "loss": 0.903,
+      "step": 4409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44362683821532e-05,
+      "loss": 0.9562,
+      "step": 4410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433808161080248e-05,
+      "loss": 0.9466,
+      "step": 4411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.443134760593147e-05,
+      "loss": 0.9477,
+      "step": 4412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428886716892268e-05,
+      "loss": 0.9289,
+      "step": 4413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442642549414806e-05,
+      "loss": 0.9392,
+      "step": 4414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423963937884293e-05,
+      "loss": 0.8796,
+      "step": 4415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4421502048286439e-05,
+      "loss": 0.9194,
+      "step": 4416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441903982554e-05,
+      "loss": 0.9377,
+      "step": 4417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441657726983049e-05,
+      "loss": 0.9349,
+      "step": 4418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414114381343458e-05,
+      "loss": 0.908,
+      "step": 4419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411651160264479e-05,
+      "loss": 0.898,
+      "step": 4420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409187606779149e-05,
+      "loss": 0.8911,
+      "step": 4421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4406723721073088e-05,
+      "loss": 0.9157,
+      "step": 4422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4404259503331948e-05,
+      "loss": 0.9116,
+      "step": 4423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4401794953741397e-05,
+      "loss": 0.9351,
+      "step": 4424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4399330072487136e-05,
+      "loss": 0.9186,
+      "step": 4425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4396864859754885e-05,
+      "loss": 0.9421,
+      "step": 4426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4394399315730389e-05,
+      "loss": 0.9218,
+      "step": 4427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4391933440599428e-05,
+      "loss": 0.962,
+      "step": 4428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4389467234547792e-05,
+      "loss": 0.9049,
+      "step": 4429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4387000697761305e-05,
+      "loss": 0.9222,
+      "step": 4430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4384533830425814e-05,
+      "loss": 0.9512,
+      "step": 4431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4382066632727192e-05,
+      "loss": 0.9243,
+      "step": 4432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379599104851336e-05,
+      "loss": 0.9023,
+      "step": 4433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377131246984168e-05,
+      "loss": 0.8951,
+      "step": 4434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4374663059311632e-05,
+      "loss": 0.9361,
+      "step": 4435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4372194542019705e-05,
+      "loss": 0.9219,
+      "step": 4436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369725695294377e-05,
+      "loss": 0.9166,
+      "step": 4437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4367256519321673e-05,
+      "loss": 0.8723,
+      "step": 4438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4364787014287636e-05,
+      "loss": 0.93,
+      "step": 4439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4362317180378337e-05,
+      "loss": 0.9762,
+      "step": 4440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4359847017779878e-05,
+      "loss": 0.879,
+      "step": 4441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357376526678368e-05,
+      "loss": 0.9403,
+      "step": 4442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354905707259962e-05,
+      "loss": 0.9705,
+      "step": 4443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352434559710826e-05,
+      "loss": 0.9332,
+      "step": 4444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349963084217154e-05,
+      "loss": 0.9567,
+      "step": 4445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4347491280965165e-05,
+      "loss": 0.8608,
+      "step": 4446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4345019150141104e-05,
+      "loss": 0.934,
+      "step": 4447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4342546691931238e-05,
+      "loss": 0.8884,
+      "step": 4448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340073906521863e-05,
+      "loss": 0.597,
+      "step": 4449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4337600794099295e-05,
+      "loss": 0.9187,
+      "step": 4450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4335127354849876e-05,
+      "loss": 0.8846,
+      "step": 4451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332653588959974e-05,
+      "loss": 0.8857,
+      "step": 4452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4330179496615984e-05,
+      "loss": 0.9597,
+      "step": 4453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4327705078004317e-05,
+      "loss": 0.9072,
+      "step": 4454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4325230333311415e-05,
+      "loss": 0.9366,
+      "step": 4455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4322755262723745e-05,
+      "loss": 0.9075,
+      "step": 4456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320279866427798e-05,
+      "loss": 0.9428,
+      "step": 4457
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4317804144610086e-05,
+      "loss": 0.9549,
+      "step": 4458
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4315328097457153e-05,
+      "loss": 0.9268,
+      "step": 4459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312851725155554e-05,
+      "loss": 0.9394,
+      "step": 4460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310375027891885e-05,
+      "loss": 0.9711,
+      "step": 4461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4307898005852756e-05,
+      "loss": 0.918,
+      "step": 4462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305420659224802e-05,
+      "loss": 0.9191,
+      "step": 4463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302942988194688e-05,
+      "loss": 0.9314,
+      "step": 4464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300464992949098e-05,
+      "loss": 0.8985,
+      "step": 4465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4297986673674747e-05,
+      "loss": 0.8927,
+      "step": 4466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4295508030558361e-05,
+      "loss": 0.8954,
+      "step": 4467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4293029063786707e-05,
+      "loss": 0.9378,
+      "step": 4468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290549773546565e-05,
+      "loss": 0.9355,
+      "step": 4469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288070160024745e-05,
+      "loss": 0.9182,
+      "step": 4470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285590223408077e-05,
+      "loss": 0.9185,
+      "step": 4471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428310996388342e-05,
+      "loss": 0.9405,
+      "step": 4472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280629381637651e-05,
+      "loss": 0.9216,
+      "step": 4473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427814847685768e-05,
+      "loss": 0.9616,
+      "step": 4474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4275667249730437e-05,
+      "loss": 0.9255,
+      "step": 4475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273185700442874e-05,
+      "loss": 0.9095,
+      "step": 4476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4270703829181969e-05,
+      "loss": 0.9201,
+      "step": 4477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4268221636134722e-05,
+      "loss": 0.9106,
+      "step": 4478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4265739121488165e-05,
+      "loss": 0.9588,
+      "step": 4479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263256285429348e-05,
+      "loss": 0.9331,
+      "step": 4480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4260773128145341e-05,
+      "loss": 0.9391,
+      "step": 4481
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258289649823252e-05,
+      "loss": 0.9614,
+      "step": 4482
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4255805850650195e-05,
+      "loss": 0.9585,
+      "step": 4483
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4253321730813326e-05,
+      "loss": 0.943,
+      "step": 4484
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425083729049981e-05,
+      "loss": 0.8769,
+      "step": 4485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4248352529896847e-05,
+      "loss": 0.9331,
+      "step": 4486
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424586744919166e-05,
+      "loss": 0.9275,
+      "step": 4487
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4243382048571485e-05,
+      "loss": 0.9261,
+      "step": 4488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4240896328223599e-05,
+      "loss": 0.8416,
+      "step": 4489
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4238410288335289e-05,
+      "loss": 0.9376,
+      "step": 4490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235923929093874e-05,
+      "loss": 0.9613,
+      "step": 4491
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4233437250686695e-05,
+      "loss": 0.9383,
+      "step": 4492
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4230950253301115e-05,
+      "loss": 0.8991,
+      "step": 4493
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4228462937124522e-05,
+      "loss": 0.9019,
+      "step": 4494
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4225975302344332e-05,
+      "loss": 0.9063,
+      "step": 4495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422348734914798e-05,
+      "loss": 0.8018,
+      "step": 4496
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4220999077722927e-05,
+      "loss": 0.9135,
+      "step": 4497
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4218510488256655e-05,
+      "loss": 0.8974,
+      "step": 4498
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4216021580936681e-05,
+      "loss": 0.912,
+      "step": 4499
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4213532355950529e-05,
+      "loss": 0.9477,
+      "step": 4500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4211042813485758e-05,
+      "loss": 0.912,
+      "step": 4501
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208552953729949e-05,
+      "loss": 0.9128,
+      "step": 4502
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206062776870707e-05,
+      "loss": 0.9331,
+      "step": 4503
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4203572283095657e-05,
+      "loss": 0.8946,
+      "step": 4504
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4201081472592455e-05,
+      "loss": 0.9242,
+      "step": 4505
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198590345548777e-05,
+      "loss": 0.9127,
+      "step": 4506
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4196098902152323e-05,
+      "loss": 0.9049,
+      "step": 4507
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4193607142590812e-05,
+      "loss": 0.936,
+      "step": 4508
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191115067051997e-05,
+      "loss": 0.8885,
+      "step": 4509
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4188622675723648e-05,
+      "loss": 0.9884,
+      "step": 4510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186129968793558e-05,
+      "loss": 0.8968,
+      "step": 4511
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183636946449547e-05,
+      "loss": 0.8983,
+      "step": 4512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.418114360887946e-05,
+      "loss": 0.9136,
+      "step": 4513
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4178649956271162e-05,
+      "loss": 0.8674,
+      "step": 4514
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417615598881254e-05,
+      "loss": 0.9852,
+      "step": 4515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173661706691514e-05,
+      "loss": 0.9463,
+      "step": 4516
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4171167110096017e-05,
+      "loss": 0.9194,
+      "step": 4517
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4168672199214011e-05,
+      "loss": 0.9639,
+      "step": 4518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166176974233483e-05,
+      "loss": 0.945,
+      "step": 4519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416368143534244e-05,
+      "loss": 0.961,
+      "step": 4520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161185582728913e-05,
+      "loss": 0.9472,
+      "step": 4521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4158689416580965e-05,
+      "loss": 0.8883,
+      "step": 4522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156192937086666e-05,
+      "loss": 0.9405,
+      "step": 4523
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4153696144434124e-05,
+      "loss": 0.9048,
+      "step": 4524
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4151199038811463e-05,
+      "loss": 0.9457,
+      "step": 4525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4148701620406838e-05,
+      "loss": 0.9376,
+      "step": 4526
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146203889408418e-05,
+      "loss": 0.938,
+      "step": 4527
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4143705846004405e-05,
+      "loss": 0.9471,
+      "step": 4528
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141207490383018e-05,
+      "loss": 0.9519,
+      "step": 4529
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4138708822732501e-05,
+      "loss": 0.9986,
+      "step": 4530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4136209843241123e-05,
+      "loss": 0.9657,
+      "step": 4531
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133710552097175e-05,
+      "loss": 0.9554,
+      "step": 4532
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131210949488972e-05,
+      "loss": 0.9555,
+      "step": 4533
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4128711035604851e-05,
+      "loss": 0.9126,
+      "step": 4534
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126210810633176e-05,
+      "loss": 0.9469,
+      "step": 4535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123710274762334e-05,
+      "loss": 0.9224,
+      "step": 4536
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121209428180731e-05,
+      "loss": 0.9709,
+      "step": 4537
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41187082710768e-05,
+      "loss": 0.908,
+      "step": 4538
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116206803638997e-05,
+      "loss": 0.9717,
+      "step": 4539
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41137050260558e-05,
+      "loss": 0.9234,
+      "step": 4540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111202938515711e-05,
+      "loss": 0.9611,
+      "step": 4541
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4108700541207259e-05,
+      "loss": 0.9259,
+      "step": 4542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4106197834318994e-05,
+      "loss": 0.9732,
+      "step": 4543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4103694818039483e-05,
+      "loss": 0.9809,
+      "step": 4544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.410119149255732e-05,
+      "loss": 0.8895,
+      "step": 4545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.409868785806113e-05,
+      "loss": 0.9224,
+      "step": 4546
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4096183914739554e-05,
+      "loss": 0.8937,
+      "step": 4547
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4093679662781256e-05,
+      "loss": 0.8847,
+      "step": 4548
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4091175102374928e-05,
+      "loss": 0.9216,
+      "step": 4549
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4088670233709278e-05,
+      "loss": 0.6035,
+      "step": 4550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4086165056973045e-05,
+      "loss": 0.8916,
+      "step": 4551
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083659572354982e-05,
+      "loss": 0.9413,
+      "step": 4552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4081153780043878e-05,
+      "loss": 0.8806,
+      "step": 4553
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407864768022853e-05,
+      "loss": 0.9533,
+      "step": 4554
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4076141273097774e-05,
+      "loss": 0.9427,
+      "step": 4555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073634558840454e-05,
+      "loss": 0.9139,
+      "step": 4556
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4071127537645453e-05,
+      "loss": 0.9514,
+      "step": 4557
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4068620209701659e-05,
+      "loss": 0.9497,
+      "step": 4558
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4066112575197996e-05,
+      "loss": 0.9492,
+      "step": 4559
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406360463432341e-05,
+      "loss": 0.9341,
+      "step": 4560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061096387266868e-05,
+      "loss": 0.9694,
+      "step": 4561
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4058587834217356e-05,
+      "loss": 0.8902,
+      "step": 4562
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4056078975363891e-05,
+      "loss": 0.9244,
+      "step": 4563
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4053569810895509e-05,
+      "loss": 0.9433,
+      "step": 4564
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051060341001264e-05,
+      "loss": 0.8988,
+      "step": 4565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404855056587024e-05,
+      "loss": 0.9328,
+      "step": 4566
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4046040485691543e-05,
+      "loss": 0.8769,
+      "step": 4567
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4043530100654301e-05,
+      "loss": 0.9359,
+      "step": 4568
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4041019410947665e-05,
+      "loss": 0.9648,
+      "step": 4569
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4038508416760808e-05,
+      "loss": 0.9008,
+      "step": 4570
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403599711828293e-05,
+      "loss": 0.5901,
+      "step": 4571
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033485515703245e-05,
+      "loss": 0.9135,
+      "step": 4572
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4030973609211001e-05,
+      "loss": 0.9174,
+      "step": 4573
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402846139899546e-05,
+      "loss": 0.919,
+      "step": 4574
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4025948885245914e-05,
+      "loss": 0.9466,
+      "step": 4575
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023436068151672e-05,
+      "loss": 0.9124,
+      "step": 4576
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4020922947902067e-05,
+      "loss": 0.9286,
+      "step": 4577
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018409524686459e-05,
+      "loss": 0.9441,
+      "step": 4578
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4015895798694224e-05,
+      "loss": 0.9197,
+      "step": 4579
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401338177011477e-05,
+      "loss": 0.9063,
+      "step": 4580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4010867439137516e-05,
+      "loss": 0.9665,
+      "step": 4581
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4008352805951917e-05,
+      "loss": 0.9173,
+      "step": 4582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4005837870747439e-05,
+      "loss": 0.9614,
+      "step": 4583
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003322633713578e-05,
+      "loss": 0.9117,
+      "step": 4584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4000807095039852e-05,
+      "loss": 0.9752,
+      "step": 4585
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3998291254915797e-05,
+      "loss": 0.9477,
+      "step": 4586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3995775113530974e-05,
+      "loss": 0.936,
+      "step": 4587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993258671074972e-05,
+      "loss": 0.9328,
+      "step": 4588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3990741927737395e-05,
+      "loss": 0.9322,
+      "step": 4589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988224883707875e-05,
+      "loss": 0.9827,
+      "step": 4590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3985707539176063e-05,
+      "loss": 0.8884,
+      "step": 4591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983189894331636e-05,
+      "loss": 0.9699,
+      "step": 4592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398067194936429e-05,
+      "loss": 0.9517,
+      "step": 4593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397815370446375e-05,
+      "loss": 0.9792,
+      "step": 4594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3975635159819757e-05,
+      "loss": 0.909,
+      "step": 4595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3973116315622068e-05,
+      "loss": 0.9845,
+      "step": 4596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970597172060482e-05,
+      "loss": 0.9464,
+      "step": 4597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3968077729324808e-05,
+      "loss": 0.8719,
+      "step": 4598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3965557987604878e-05,
+      "loss": 0.9126,
+      "step": 4599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396303794709055e-05,
+      "loss": 0.956,
+      "step": 4600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3960517607971697e-05,
+      "loss": 0.9598,
+      "step": 4601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957996970438226e-05,
+      "loss": 0.9427,
+      "step": 4602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955476034680058e-05,
+      "loss": 0.9258,
+      "step": 4603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3952954800887137e-05,
+      "loss": 0.5689,
+      "step": 4604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3950433269249436e-05,
+      "loss": 0.8942,
+      "step": 4605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3947911439956942e-05,
+      "loss": 0.9447,
+      "step": 4606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3945389313199669e-05,
+      "loss": 0.9629,
+      "step": 4607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3942866889167653e-05,
+      "loss": 0.9203,
+      "step": 4608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940344168050952e-05,
+      "loss": 0.9512,
+      "step": 4609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3937821150039645e-05,
+      "loss": 0.8606,
+      "step": 4610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3935297835323835e-05,
+      "loss": 0.9175,
+      "step": 4611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932774224093653e-05,
+      "loss": 0.928,
+      "step": 4612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930250316539237e-05,
+      "loss": 0.9278,
+      "step": 4613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927726112850764e-05,
+      "loss": 0.9225,
+      "step": 4614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3925201613218424e-05,
+      "loss": 0.9146,
+      "step": 4615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3922676817832432e-05,
+      "loss": 0.942,
+      "step": 4616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3920151726883021e-05,
+      "loss": 0.9349,
+      "step": 4617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3917626340560454e-05,
+      "loss": 0.9099,
+      "step": 4618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391510065905501e-05,
+      "loss": 0.9038,
+      "step": 4619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3912574682556994e-05,
+      "loss": 0.8909,
+      "step": 4620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910048411256734e-05,
+      "loss": 0.8848,
+      "step": 4621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3907521845344571e-05,
+      "loss": 0.8502,
+      "step": 4622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3904994985010882e-05,
+      "loss": 0.8884,
+      "step": 4623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3902467830446056e-05,
+      "loss": 0.9178,
+      "step": 4624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.389994038184051e-05,
+      "loss": 0.9274,
+      "step": 4625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3897412639384675e-05,
+      "loss": 0.8977,
+      "step": 4626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894884603269016e-05,
+      "loss": 0.9066,
+      "step": 4627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3892356273684012e-05,
+      "loss": 0.9099,
+      "step": 4628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3889827650820165e-05,
+      "loss": 0.9547,
+      "step": 4629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887298734868e-05,
+      "loss": 0.9065,
+      "step": 4630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884769526018063e-05,
+      "loss": 0.9309,
+      "step": 4631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3882240024460928e-05,
+      "loss": 0.8962,
+      "step": 4632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387971023038718e-05,
+      "loss": 0.9613,
+      "step": 4633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3877180143987436e-05,
+      "loss": 0.9512,
+      "step": 4634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3874649765452333e-05,
+      "loss": 0.9152,
+      "step": 4635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3872119094972525e-05,
+      "loss": 0.9367,
+      "step": 4636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3869588132738696e-05,
+      "loss": 0.9586,
+      "step": 4637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386705687894154e-05,
+      "loss": 0.9386,
+      "step": 4638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3864525333771787e-05,
+      "loss": 0.5285,
+      "step": 4639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861993497420176e-05,
+      "loss": 0.9244,
+      "step": 4640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385946137007748e-05,
+      "loss": 0.9263,
+      "step": 4641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3856928951934487e-05,
+      "loss": 0.9662,
+      "step": 4642
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3854396243182007e-05,
+      "loss": 0.9225,
+      "step": 4643
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3851863244010872e-05,
+      "loss": 0.5329,
+      "step": 4644
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3849329954611941e-05,
+      "loss": 0.9014,
+      "step": 4645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846796375176083e-05,
+      "loss": 0.872,
+      "step": 4646
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844262505894205e-05,
+      "loss": 0.9106,
+      "step": 4647
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3841728346957226e-05,
+      "loss": 0.8679,
+      "step": 4648
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3839193898556083e-05,
+      "loss": 0.8781,
+      "step": 4649
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836659160881742e-05,
+      "loss": 0.9335,
+      "step": 4650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3834124134125191e-05,
+      "loss": 0.9009,
+      "step": 4651
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831588818477437e-05,
+      "loss": 0.9006,
+      "step": 4652
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3829053214129506e-05,
+      "loss": 0.9301,
+      "step": 4653
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826517321272457e-05,
+      "loss": 0.9338,
+      "step": 4654
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3823981140097354e-05,
+      "loss": 0.9595,
+      "step": 4655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3821444670795296e-05,
+      "loss": 0.8911,
+      "step": 4656
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38189079135574e-05,
+      "loss": 0.9176,
+      "step": 4657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816370868574804e-05,
+      "loss": 0.9434,
+      "step": 4658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3813833536038663e-05,
+      "loss": 0.8758,
+      "step": 4659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3811295916140164e-05,
+      "loss": 0.9509,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3808758009070506e-05,
+      "loss": 0.9512,
+      "step": 4661
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3806219815020916e-05,
+      "loss": 0.9365,
+      "step": 4662
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3803681334182639e-05,
+      "loss": 0.9218,
+      "step": 4663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3801142566746945e-05,
+      "loss": 0.9078,
+      "step": 4664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3798603512905123e-05,
+      "loss": 0.9346,
+      "step": 4665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3796064172848482e-05,
+      "loss": 0.966,
+      "step": 4666
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3793524546768358e-05,
+      "loss": 0.9723,
+      "step": 4667
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37909846348561e-05,
+      "loss": 0.9358,
+      "step": 4668
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378844443730309e-05,
+      "loss": 0.885,
+      "step": 4669
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378590395430072e-05,
+      "loss": 0.9447,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378336318604041e-05,
+      "loss": 0.944,
+      "step": 4671
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37808221327136e-05,
+      "loss": 0.9242,
+      "step": 4672
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3778280794511753e-05,
+      "loss": 0.8886,
+      "step": 4673
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3775739171626354e-05,
+      "loss": 0.8756,
+      "step": 4674
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3773197264248903e-05,
+      "loss": 0.5488,
+      "step": 4675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770655072570929e-05,
+      "loss": 0.932,
+      "step": 4676
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376811259678398e-05,
+      "loss": 0.9439,
+      "step": 4677
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3765569837079622e-05,
+      "loss": 0.9464,
+      "step": 4678
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376302679364945e-05,
+      "loss": 0.921,
+      "step": 4679
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376048346668507e-05,
+      "loss": 0.9405,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375793985637812e-05,
+      "loss": 0.9129,
+      "step": 4681
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375539596292025e-05,
+      "loss": 0.9752,
+      "step": 4682
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3752851786503139e-05,
+      "loss": 0.9383,
+      "step": 4683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375030732731848e-05,
+      "loss": 0.9523,
+      "step": 4684
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747762585557995e-05,
+      "loss": 0.9311,
+      "step": 4685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745217561413424e-05,
+      "loss": 0.9099,
+      "step": 4686
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742672255076528e-05,
+      "loss": 0.6066,
+      "step": 4687
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740126666739086e-05,
+      "loss": 0.9084,
+      "step": 4688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3737580796592905e-05,
+      "loss": 0.9996,
+      "step": 4689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3735034644829806e-05,
+      "loss": 0.8892,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732488211641638e-05,
+      "loss": 0.9489,
+      "step": 4691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729941497220267e-05,
+      "loss": 0.9277,
+      "step": 4692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3727394501757578e-05,
+      "loss": 0.9224,
+      "step": 4693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3724847225445488e-05,
+      "loss": 0.9033,
+      "step": 4694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372229966847592e-05,
+      "loss": 0.9381,
+      "step": 4695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719751831040831e-05,
+      "loss": 0.877,
+      "step": 4696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371720371333219e-05,
+      "loss": 0.9985,
+      "step": 4697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714655315541992e-05,
+      "loss": 0.9385,
+      "step": 4698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712106637862254e-05,
+      "loss": 0.954,
+      "step": 4699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709557680485013e-05,
+      "loss": 0.9055,
+      "step": 4700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3707008443602323e-05,
+      "loss": 0.9355,
+      "step": 4701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704458927406261e-05,
+      "loss": 0.9306,
+      "step": 4702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701909132088931e-05,
+      "loss": 0.9464,
+      "step": 4703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3699359057842451e-05,
+      "loss": 0.9461,
+      "step": 4704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696808704858964e-05,
+      "loss": 0.9578,
+      "step": 4705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3694258073330626e-05,
+      "loss": 0.9127,
+      "step": 4706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3691707163449631e-05,
+      "loss": 0.9506,
+      "step": 4707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689155975408174e-05,
+      "loss": 0.9136,
+      "step": 4708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3686604509398489e-05,
+      "loss": 0.9261,
+      "step": 4709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684052765612814e-05,
+      "loss": 0.9441,
+      "step": 4710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3681500744243423e-05,
+      "loss": 0.9763,
+      "step": 4711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678948445482598e-05,
+      "loss": 0.917,
+      "step": 4712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676395869522653e-05,
+      "loss": 0.9466,
+      "step": 4713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673843016555912e-05,
+      "loss": 0.8947,
+      "step": 4714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3671289886774733e-05,
+      "loss": 0.9304,
+      "step": 4715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3668736480371484e-05,
+      "loss": 0.95,
+      "step": 4716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3666182797538554e-05,
+      "loss": 0.9664,
+      "step": 4717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3663628838468364e-05,
+      "loss": 0.9425,
+      "step": 4718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3661074603353341e-05,
+      "loss": 0.8785,
+      "step": 4719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3658520092385946e-05,
+      "loss": 0.8958,
+      "step": 4720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655965305758652e-05,
+      "loss": 0.95,
+      "step": 4721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 0.905,
+      "step": 4722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3650854906294368e-05,
+      "loss": 0.9888,
+      "step": 4723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3648299293842438e-05,
+      "loss": 0.6073,
+      "step": 4724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645743406500715e-05,
+      "loss": 0.9103,
+      "step": 4725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3643187244461787e-05,
+      "loss": 0.9254,
+      "step": 4726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3640630807918246e-05,
+      "loss": 0.9363,
+      "step": 4727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3638074097062716e-05,
+      "loss": 0.9012,
+      "step": 4728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3635517112087841e-05,
+      "loss": 0.9613,
+      "step": 4729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363295985318628e-05,
+      "loss": 0.8937,
+      "step": 4730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363040232055072e-05,
+      "loss": 0.9235,
+      "step": 4731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362784451437386e-05,
+      "loss": 0.934,
+      "step": 4732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3625286434848424e-05,
+      "loss": 0.9223,
+      "step": 4733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622728082167159e-05,
+      "loss": 0.9471,
+      "step": 4734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3620169456522824e-05,
+      "loss": 0.8824,
+      "step": 4735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617610558108214e-05,
+      "loss": 0.9254,
+      "step": 4736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3615051387116131e-05,
+      "loss": 0.8912,
+      "step": 4737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3612491943739401e-05,
+      "loss": 0.9362,
+      "step": 4738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609932228170873e-05,
+      "loss": 0.8903,
+      "step": 4739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3607372240603411e-05,
+      "loss": 0.9307,
+      "step": 4740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3604811981229906e-05,
+      "loss": 0.897,
+      "step": 4741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602251450243273e-05,
+      "loss": 0.878,
+      "step": 4742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359969064783643e-05,
+      "loss": 0.9433,
+      "step": 4743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3597129574202334e-05,
+      "loss": 0.8988,
+      "step": 4744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3594568229533953e-05,
+      "loss": 0.8656,
+      "step": 4745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3592006614024278e-05,
+      "loss": 0.8785,
+      "step": 4746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358944472786632e-05,
+      "loss": 0.9152,
+      "step": 4747
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358688257125311e-05,
+      "loss": 0.9626,
+      "step": 4748
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35843201443777e-05,
+      "loss": 0.9397,
+      "step": 4749
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3581757447433162e-05,
+      "loss": 0.9035,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357919448061259e-05,
+      "loss": 0.9587,
+      "step": 4751
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3576631244109093e-05,
+      "loss": 0.9283,
+      "step": 4752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3574067738115812e-05,
+      "loss": 0.939,
+      "step": 4753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571503962825892e-05,
+      "loss": 0.9573,
+      "step": 4754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.356893991843251e-05,
+      "loss": 0.9563,
+      "step": 4755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566375605128861e-05,
+      "loss": 0.938,
+      "step": 4756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3563811023108157e-05,
+      "loss": 0.9191,
+      "step": 4757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3561246172563633e-05,
+      "loss": 0.8995,
+      "step": 4758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3558681053688547e-05,
+      "loss": 0.9388,
+      "step": 4759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355611566667617e-05,
+      "loss": 0.5216,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35535500117198e-05,
+      "loss": 0.9287,
+      "step": 4761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355098408901275e-05,
+      "loss": 0.9126,
+      "step": 4762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3548417898748361e-05,
+      "loss": 0.9358,
+      "step": 4763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354585144111998e-05,
+      "loss": 0.9041,
+      "step": 4764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3543284716320989e-05,
+      "loss": 0.9377,
+      "step": 4765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354071772454478e-05,
+      "loss": 0.9411,
+      "step": 4766
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3538150465984771e-05,
+      "loss": 0.8596,
+      "step": 4767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35355829408344e-05,
+      "loss": 0.9266,
+      "step": 4768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3533015149287123e-05,
+      "loss": 0.9506,
+      "step": 4769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530447091536413e-05,
+      "loss": 0.9097,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.352787876777577e-05,
+      "loss": 0.958,
+      "step": 4771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3525310178198707e-05,
+      "loss": 0.8672,
+      "step": 4772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3522741322998764e-05,
+      "loss": 0.9267,
+      "step": 4773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35201722023695e-05,
+      "loss": 0.9234,
+      "step": 4774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3517602816504482e-05,
+      "loss": 0.8798,
+      "step": 4775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3515033165597315e-05,
+      "loss": 0.921,
+      "step": 4776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3512463249841613e-05,
+      "loss": 0.9186,
+      "step": 4777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3509893069431011e-05,
+      "loss": 0.9137,
+      "step": 4778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3507322624559165e-05,
+      "loss": 0.892,
+      "step": 4779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3504751915419754e-05,
+      "loss": 0.9361,
+      "step": 4780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3502180942206472e-05,
+      "loss": 0.9289,
+      "step": 4781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3499609705113034e-05,
+      "loss": 0.9361,
+      "step": 4782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3497038204333182e-05,
+      "loss": 0.923,
+      "step": 4783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3494466440060667e-05,
+      "loss": 0.9581,
+      "step": 4784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3491894412489264e-05,
+      "loss": 0.9317,
+      "step": 4785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3489322121812772e-05,
+      "loss": 0.9577,
+      "step": 4786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3486749568225002e-05,
+      "loss": 0.9366,
+      "step": 4787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348417675191979e-05,
+      "loss": 0.9433,
+      "step": 4788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3481603673090994e-05,
+      "loss": 0.9117,
+      "step": 4789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3479030331932488e-05,
+      "loss": 0.943,
+      "step": 4790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476456728638168e-05,
+      "loss": 0.931,
+      "step": 4791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473882863401941e-05,
+      "loss": 0.8626,
+      "step": 4792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347130873641775e-05,
+      "loss": 0.9189,
+      "step": 4793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346873434787954e-05,
+      "loss": 0.9448,
+      "step": 4794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3466159697981294e-05,
+      "loss": 0.9673,
+      "step": 4795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463584786916997e-05,
+      "loss": 0.9559,
+      "step": 4796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461009614880668e-05,
+      "loss": 0.606,
+      "step": 4797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458434182066333e-05,
+      "loss": 0.9373,
+      "step": 4798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345585848866805e-05,
+      "loss": 0.8809,
+      "step": 4799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3453282534879888e-05,
+      "loss": 0.9418,
+      "step": 4800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450706320895938e-05,
+      "loss": 0.9277,
+      "step": 4801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448129846910312e-05,
+      "loss": 0.9407,
+      "step": 4802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344555311311714e-05,
+      "loss": 0.9363,
+      "step": 4803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442976119710571e-05,
+      "loss": 0.8941,
+      "step": 4804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3440398866884781e-05,
+      "loss": 0.9335,
+      "step": 4805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3437821354833951e-05,
+      "loss": 0.9211,
+      "step": 4806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435243583752294e-05,
+      "loss": 0.8966,
+      "step": 4807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3432665553834036e-05,
+      "loss": 0.9082,
+      "step": 4808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430087265273426e-05,
+      "loss": 0.8894,
+      "step": 4809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427508718264732e-05,
+      "loss": 0.9563,
+      "step": 4810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342492991300224e-05,
+      "loss": 0.89,
+      "step": 4811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3422350849680258e-05,
+      "loss": 0.9358,
+      "step": 4812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3419771528493108e-05,
+      "loss": 0.9692,
+      "step": 4813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417191949635137e-05,
+      "loss": 0.9383,
+      "step": 4814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3414612113300714e-05,
+      "loss": 0.9222,
+      "step": 4815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3412032019684213e-05,
+      "loss": 0.9254,
+      "step": 4816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3409451668980047e-05,
+      "loss": 0.9882,
+      "step": 4817
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3406871061382632e-05,
+      "loss": 0.9107,
+      "step": 4818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3404290197086414e-05,
+      "loss": 0.9253,
+      "step": 4819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3401709076285854e-05,
+      "loss": 0.9,
+      "step": 4820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3399127699175429e-05,
+      "loss": 0.9186,
+      "step": 4821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3396546065949643e-05,
+      "loss": 0.9398,
+      "step": 4822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393964176803014e-05,
+      "loss": 0.9252,
+      "step": 4823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339138203193008e-05,
+      "loss": 0.5643,
+      "step": 4824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33887996315254e-05,
+      "loss": 0.9323,
+      "step": 4825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338621697578355e-05,
+      "loss": 0.5746,
+      "step": 4826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3383634064899127e-05,
+      "loss": 0.9284,
+      "step": 4827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381050899066748e-05,
+      "loss": 0.9487,
+      "step": 4828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3378467478481043e-05,
+      "loss": 0.8995,
+      "step": 4829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3375883803336671e-05,
+      "loss": 0.9494,
+      "step": 4830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3373299873828303e-05,
+      "loss": 0.9419,
+      "step": 4831
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370715690150631e-05,
+      "loss": 0.9512,
+      "step": 4832
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3368131252498366e-05,
+      "loss": 0.9222,
+      "step": 4833
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3365546561066242e-05,
+      "loss": 0.8913,
+      "step": 4834
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3362961616049006e-05,
+      "loss": 0.92,
+      "step": 4835
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3360376417641425e-05,
+      "loss": 0.9108,
+      "step": 4836
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3357790966038292e-05,
+      "loss": 0.9137,
+      "step": 4837
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3355205261434408e-05,
+      "loss": 0.9289,
+      "step": 4838
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3352619304024606e-05,
+      "loss": 0.954,
+      "step": 4839
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3350033094003724e-05,
+      "loss": 0.971,
+      "step": 4840
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334744663156663e-05,
+      "loss": 0.8958,
+      "step": 4841
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3344859916908206e-05,
+      "loss": 0.9468,
+      "step": 4842
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342272950223359e-05,
+      "loss": 0.939,
+      "step": 4843
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339685731707002e-05,
+      "loss": 0.9297,
+      "step": 4844
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333709826155408e-05,
+      "loss": 0.9424,
+      "step": 4845
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3334510539959551e-05,
+      "loss": 0.9136,
+      "step": 4846
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3331922567118394e-05,
+      "loss": 0.9365,
+      "step": 4847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329334343225607e-05,
+      "loss": 0.9311,
+      "step": 4848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33267458684762e-05,
+      "loss": 0.948,
+      "step": 4849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324157143065213e-05,
+      "loss": 0.9045,
+      "step": 4850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3321568167187699e-05,
+      "loss": 0.9481,
+      "step": 4851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331897894103873e-05,
+      "loss": 0.879,
+      "step": 4852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3316389464813397e-05,
+      "loss": 0.8966,
+      "step": 4853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3313799738706812e-05,
+      "loss": 0.8814,
+      "step": 4854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3311209762914104e-05,
+      "loss": 0.9255,
+      "step": 4855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308619537630416e-05,
+      "loss": 0.9334,
+      "step": 4856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330602906305092e-05,
+      "loss": 0.911,
+      "step": 4857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33034383393708e-05,
+      "loss": 0.908,
+      "step": 4858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300847366785261e-05,
+      "loss": 0.905,
+      "step": 4859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3298256145489524e-05,
+      "loss": 0.9056,
+      "step": 4860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329566467567883e-05,
+      "loss": 0.9321,
+      "step": 4861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3293072957548443e-05,
+      "loss": 0.8875,
+      "step": 4862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3290480991293642e-05,
+      "loss": 0.9267,
+      "step": 4863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3287888777109721e-05,
+      "loss": 0.8802,
+      "step": 4864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285296315192e-05,
+      "loss": 0.8608,
+      "step": 4865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3282703605735811e-05,
+      "loss": 0.9281,
+      "step": 4866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280110648936512e-05,
+      "loss": 0.8969,
+      "step": 4867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3277517444989476e-05,
+      "loss": 0.9212,
+      "step": 4868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3274923994090088e-05,
+      "loss": 0.8921,
+      "step": 4869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3272330296433761e-05,
+      "loss": 0.9338,
+      "step": 4870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3269736352215925e-05,
+      "loss": 0.9175,
+      "step": 4871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3267142161632025e-05,
+      "loss": 0.9211,
+      "step": 4872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3264547724877525e-05,
+      "loss": 0.9585,
+      "step": 4873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3261953042147915e-05,
+      "loss": 0.9228,
+      "step": 4874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325935811363869e-05,
+      "loss": 0.8718,
+      "step": 4875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256762939545375e-05,
+      "loss": 0.9237,
+      "step": 4876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325416752006351e-05,
+      "loss": 0.9294,
+      "step": 4877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251571855388652e-05,
+      "loss": 0.8912,
+      "step": 4878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248975945716378e-05,
+      "loss": 0.9546,
+      "step": 4879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246379791242284e-05,
+      "loss": 0.9082,
+      "step": 4880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3243783392161977e-05,
+      "loss": 0.9148,
+      "step": 4881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32411867486711e-05,
+      "loss": 0.9275,
+      "step": 4882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3238589860965295e-05,
+      "loss": 0.9136,
+      "step": 4883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3235992729240232e-05,
+      "loss": 0.973,
+      "step": 4884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233395353691598e-05,
+      "loss": 0.944,
+      "step": 4885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3230797734515102e-05,
+      "loss": 0.9304,
+      "step": 4886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228199871906465e-05,
+      "loss": 0.957,
+      "step": 4887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322560176606143e-05,
+      "loss": 0.9565,
+      "step": 4888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3223003417175755e-05,
+      "loss": 0.9193,
+      "step": 4889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322040482544522e-05,
+      "loss": 0.9459,
+      "step": 4890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217805991065624e-05,
+      "loss": 0.9183,
+      "step": 4891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.321520691423278e-05,
+      "loss": 0.9167,
+      "step": 4892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3212607595142525e-05,
+      "loss": 0.8745,
+      "step": 4893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3210008033990706e-05,
+      "loss": 0.9536,
+      "step": 4894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3207408230973198e-05,
+      "loss": 0.8929,
+      "step": 4895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3204808186285884e-05,
+      "loss": 0.9242,
+      "step": 4896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3202207900124677e-05,
+      "loss": 0.9306,
+      "step": 4897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3199607372685497e-05,
+      "loss": 0.9401,
+      "step": 4898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3197006604164289e-05,
+      "loss": 0.9552,
+      "step": 4899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3194405594757012e-05,
+      "loss": 0.9484,
+      "step": 4900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3191804344659647e-05,
+      "loss": 0.9143,
+      "step": 4901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3191804344659647e-05,
+      "loss": 0.917,
+      "step": 4902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.318920285406819e-05,
+      "loss": 0.9529,
+      "step": 4903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3186601123178658e-05,
+      "loss": 0.9029,
+      "step": 4904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3183999152187084e-05,
+      "loss": 0.914,
+      "step": 4905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3181396941289521e-05,
+      "loss": 0.9231,
+      "step": 4906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3178794490682036e-05,
+      "loss": 0.9228,
+      "step": 4907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317619180056072e-05,
+      "loss": 0.9326,
+      "step": 4908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173588871121674e-05,
+      "loss": 0.9377,
+      "step": 4909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317098570256103e-05,
+      "loss": 0.9253,
+      "step": 4910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3168382295074923e-05,
+      "loss": 0.9572,
+      "step": 4911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165778648859515e-05,
+      "loss": 0.9374,
+      "step": 4912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3163174764110985e-05,
+      "loss": 0.9624,
+      "step": 4913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3160570641025526e-05,
+      "loss": 0.9649,
+      "step": 4914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157966279799352e-05,
+      "loss": 0.9076,
+      "step": 4915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3155361680628698e-05,
+      "loss": 0.5757,
+      "step": 4916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3152756843709814e-05,
+      "loss": 0.9588,
+      "step": 4917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3150151769238964e-05,
+      "loss": 0.9363,
+      "step": 4918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3147546457412435e-05,
+      "loss": 0.8886,
+      "step": 4919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144940908426532e-05,
+      "loss": 0.9373,
+      "step": 4920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3142335122477575e-05,
+      "loss": 0.9018,
+      "step": 4921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31397290997619e-05,
+      "loss": 0.9107,
+      "step": 4922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313712284047587e-05,
+      "loss": 0.9289,
+      "step": 4923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3134516344815857e-05,
+      "loss": 0.9521,
+      "step": 4924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131909612978254e-05,
+      "loss": 0.5245,
+      "step": 4925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.312930264515947e-05,
+      "loss": 0.9103,
+      "step": 4926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3126695441555935e-05,
+      "loss": 0.9325,
+      "step": 4927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3124088002364093e-05,
+      "loss": 0.9492,
+      "step": 4928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121480327780409e-05,
+      "loss": 0.9488,
+      "step": 4929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3118872418001365e-05,
+      "loss": 0.9347,
+      "step": 4930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3116264273223459e-05,
+      "loss": 0.9213,
+      "step": 4931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3113655893643208e-05,
+      "loss": 0.9386,
+      "step": 4932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311104727945715e-05,
+      "loss": 0.8964,
+      "step": 4933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3108438430861832e-05,
+      "loss": 0.9254,
+      "step": 4934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105829348053824e-05,
+      "loss": 0.9661,
+      "step": 4935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3103220031229717e-05,
+      "loss": 0.8953,
+      "step": 4936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100610480586115e-05,
+      "loss": 0.9171,
+      "step": 4937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3098000696319642e-05,
+      "loss": 0.9233,
+      "step": 4938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095390678626935e-05,
+      "loss": 0.9052,
+      "step": 4939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3092780427704655e-05,
+      "loss": 0.9023,
+      "step": 4940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.9191,
+      "step": 4941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3087559226958092e-05,
+      "loss": 0.9095,
+      "step": 4942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084948277527211e-05,
+      "loss": 0.9415,
+      "step": 4943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3082337095653569e-05,
+      "loss": 0.8753,
+      "step": 4944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3079725681533902e-05,
+      "loss": 0.9114,
+      "step": 4945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3077114035364978e-05,
+      "loss": 0.9625,
+      "step": 4946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3074502157343575e-05,
+      "loss": 0.9711,
+      "step": 4947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3071890047666498e-05,
+      "loss": 0.8443,
+      "step": 4948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3069277706530555e-05,
+      "loss": 0.9282,
+      "step": 4949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3066665134132584e-05,
+      "loss": 0.8553,
+      "step": 4950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306405233066943e-05,
+      "loss": 0.9566,
+      "step": 4951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3061439296337968e-05,
+      "loss": 0.8968,
+      "step": 4952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305882603133508e-05,
+      "loss": 0.9065,
+      "step": 4953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3056212535857666e-05,
+      "loss": 0.962,
+      "step": 4954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3053598810102647e-05,
+      "loss": 0.9022,
+      "step": 4955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3050984854266963e-05,
+      "loss": 0.9076,
+      "step": 4956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304837066854757e-05,
+      "loss": 0.9155,
+      "step": 4957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3045756253141435e-05,
+      "loss": 0.8679,
+      "step": 4958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3043141608245551e-05,
+      "loss": 0.9257,
+      "step": 4959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040526734056922e-05,
+      "loss": 0.8625,
+      "step": 4960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3037911630772576e-05,
+      "loss": 0.9045,
+      "step": 4961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3035296298589549e-05,
+      "loss": 0.9044,
+      "step": 4962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032680737704902e-05,
+      "loss": 0.8912,
+      "step": 4963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303006494831571e-05,
+      "loss": 0.9012,
+      "step": 4964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.302744893061907e-05,
+      "loss": 0.8932,
+      "step": 4965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3024832684812086e-05,
+      "loss": 0.8945,
+      "step": 4966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022216211091887e-05,
+      "loss": 0.9711,
+      "step": 4967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301959950965562e-05,
+      "loss": 0.5354,
+      "step": 4968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016982580700444e-05,
+      "loss": 0.8967,
+      "step": 4969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3014365424423539e-05,
+      "loss": 0.9731,
+      "step": 4970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011748041022101e-05,
+      "loss": 0.9296,
+      "step": 4971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3009130430693342e-05,
+      "loss": 0.9183,
+      "step": 4972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3006512593634492e-05,
+      "loss": 0.5537,
+      "step": 4973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3003894530042803e-05,
+      "loss": 0.6401,
+      "step": 4974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300127624011553e-05,
+      "loss": 0.942,
+      "step": 4975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299865772404996e-05,
+      "loss": 0.5824,
+      "step": 4976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299603898204339e-05,
+      "loss": 0.9164,
+      "step": 4977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993420014293139e-05,
+      "loss": 0.9237,
+      "step": 4978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990800820996534e-05,
+      "loss": 0.9164,
+      "step": 4979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988181402350926e-05,
+      "loss": 0.9161,
+      "step": 4980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2985561758553683e-05,
+      "loss": 0.8998,
+      "step": 4981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982941889802188e-05,
+      "loss": 0.8722,
+      "step": 4982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2980321796293838e-05,
+      "loss": 0.894,
+      "step": 4983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2977701478226054e-05,
+      "loss": 0.8963,
+      "step": 4984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297508093579627e-05,
+      "loss": 0.902,
+      "step": 4985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2972460169201933e-05,
+      "loss": 0.9591,
+      "step": 4986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969839178640515e-05,
+      "loss": 0.9437,
+      "step": 4987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29672179643095e-05,
+      "loss": 0.9128,
+      "step": 4988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296459652640639e-05,
+      "loss": 0.9095,
+      "step": 4989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29619748651287e-05,
+      "loss": 0.9378,
+      "step": 4990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2959352980673969e-05,
+      "loss": 0.929,
+      "step": 4991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2956730873239746e-05,
+      "loss": 0.9641,
+      "step": 4992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954108543023605e-05,
+      "loss": 0.922,
+      "step": 4993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2951485990223127e-05,
+      "loss": 0.9412,
+      "step": 4994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2948863215035918e-05,
+      "loss": 0.9227,
+      "step": 4995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2946240217659596e-05,
+      "loss": 0.9234,
+      "step": 4996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2943616998291793e-05,
+      "loss": 0.9172,
+      "step": 4997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940993557130166e-05,
+      "loss": 0.9067,
+      "step": 4998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2938369894372382e-05,
+      "loss": 0.8835,
+      "step": 4999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293574601021613e-05,
+      "loss": 0.9172,
+      "step": 5000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933121904859111e-05,
+      "loss": 0.8687,
+      "step": 5001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2930497578499047e-05,
+      "loss": 0.9552,
+      "step": 5002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927873031333666e-05,
+      "loss": 0.9018,
+      "step": 5003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2925248263560733e-05,
+      "loss": 0.9168,
+      "step": 5004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2922623275378011e-05,
+      "loss": 0.6258,
+      "step": 5005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919998066983279e-05,
+      "loss": 0.9427,
+      "step": 5006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291737263857435e-05,
+      "loss": 0.9091,
+      "step": 5007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2914746990349041e-05,
+      "loss": 0.9798,
+      "step": 5008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2912121122505186e-05,
+      "loss": 0.8796,
+      "step": 5009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2909495035240638e-05,
+      "loss": 0.9089,
+      "step": 5010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906868728753264e-05,
+      "loss": 0.9491,
+      "step": 5011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290424220324095e-05,
+      "loss": 0.9371,
+      "step": 5012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2901615458901602e-05,
+      "loss": 0.9827,
+      "step": 5013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2898988495933132e-05,
+      "loss": 0.9589,
+      "step": 5014
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2896361314533478e-05,
+      "loss": 0.5648,
+      "step": 5015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2893733914900595e-05,
+      "loss": 0.9202,
+      "step": 5016
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2891106297232442e-05,
+      "loss": 0.942,
+      "step": 5017
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.288847846172701e-05,
+      "loss": 0.8869,
+      "step": 5018
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2885850408582295e-05,
+      "loss": 0.8878,
+      "step": 5019
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2883222137996316e-05,
+      "loss": 0.9676,
+      "step": 5020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2880593650167107e-05,
+      "loss": 0.9837,
+      "step": 5021
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2877964945292717e-05,
+      "loss": 0.9039,
+      "step": 5022
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2875336023571215e-05,
+      "loss": 0.8938,
+      "step": 5023
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2872706885200676e-05,
+      "loss": 0.8949,
+      "step": 5024
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2870077530379205e-05,
+      "loss": 0.925,
+      "step": 5025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2867447959304918e-05,
+      "loss": 0.9262,
+      "step": 5026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864818172175938e-05,
+      "loss": 0.8921,
+      "step": 5027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2862188169190419e-05,
+      "loss": 0.8891,
+      "step": 5028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859557950546527e-05,
+      "loss": 0.9342,
+      "step": 5029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2856927516442435e-05,
+      "loss": 0.9543,
+      "step": 5030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2854296867076346e-05,
+      "loss": 0.9661,
+      "step": 5031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2851666002646468e-05,
+      "loss": 0.9154,
+      "step": 5032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849034923351034e-05,
+      "loss": 0.9167,
+      "step": 5033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846403629388285e-05,
+      "loss": 0.9046,
+      "step": 5034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2843772120956483e-05,
+      "loss": 0.9194,
+      "step": 5035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841140398253907e-05,
+      "loss": 0.8946,
+      "step": 5036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283850846147885e-05,
+      "loss": 0.9229,
+      "step": 5037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2835876310829619e-05,
+      "loss": 0.9545,
+      "step": 5038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2833243946504541e-05,
+      "loss": 0.8873,
+      "step": 5039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2830611368701957e-05,
+      "loss": 0.9222,
+      "step": 5040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2827978577620228e-05,
+      "loss": 0.9332,
+      "step": 5041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825345573457723e-05,
+      "loss": 0.925,
+      "step": 5042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282271235641284e-05,
+      "loss": 0.9603,
+      "step": 5043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2820078926683976e-05,
+      "loss": 0.8738,
+      "step": 5044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2817445284469556e-05,
+      "loss": 0.9216,
+      "step": 5045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2814811429968022e-05,
+      "loss": 0.9474,
+      "step": 5046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2812177363377825e-05,
+      "loss": 0.8679,
+      "step": 5047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2809543084897434e-05,
+      "loss": 0.954,
+      "step": 5048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2806908594725335e-05,
+      "loss": 0.8834,
+      "step": 5049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2804273893060028e-05,
+      "loss": 0.9139,
+      "step": 5050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2801638980100038e-05,
+      "loss": 0.9036,
+      "step": 5051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799003856043893e-05,
+      "loss": 0.9045,
+      "step": 5052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796368521090143e-05,
+      "loss": 0.8985,
+      "step": 5053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2793732975437357e-05,
+      "loss": 0.9082,
+      "step": 5054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2791097219284115e-05,
+      "loss": 0.8905,
+      "step": 5055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2788461252829009e-05,
+      "loss": 0.9199,
+      "step": 5056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2785825076270662e-05,
+      "loss": 0.8824,
+      "step": 5057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2783188689807697e-05,
+      "loss": 0.9052,
+      "step": 5058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780552093638757e-05,
+      "loss": 0.9412,
+      "step": 5059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777915287962506e-05,
+      "loss": 0.917,
+      "step": 5060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277527827297762e-05,
+      "loss": 0.9026,
+      "step": 5061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772641048882787e-05,
+      "loss": 0.9241,
+      "step": 5062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2770003615876725e-05,
+      "loss": 0.9677,
+      "step": 5063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767365974158146e-05,
+      "loss": 0.8943,
+      "step": 5064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2764728123925796e-05,
+      "loss": 0.9283,
+      "step": 5065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276209006537843e-05,
+      "loss": 0.8518,
+      "step": 5066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2759451798714816e-05,
+      "loss": 0.9287,
+      "step": 5067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2756813324133746e-05,
+      "loss": 0.9189,
+      "step": 5068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2754174641834014e-05,
+      "loss": 0.9561,
+      "step": 5069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2751535752014444e-05,
+      "loss": 0.9341,
+      "step": 5070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748896654873865e-05,
+      "loss": 0.9586,
+      "step": 5071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2746257350611129e-05,
+      "loss": 0.9705,
+      "step": 5072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27436178394251e-05,
+      "loss": 0.952,
+      "step": 5073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2740978121514659e-05,
+      "loss": 0.9256,
+      "step": 5074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27383381970787e-05,
+      "loss": 0.9403,
+      "step": 5075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2735698066316138e-05,
+      "loss": 0.9494,
+      "step": 5076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2733057729425898e-05,
+      "loss": 0.9755,
+      "step": 5077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273041718660692e-05,
+      "loss": 0.9148,
+      "step": 5078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727776438058166e-05,
+      "loss": 0.9258,
+      "step": 5079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2725135483978606e-05,
+      "loss": 0.9086,
+      "step": 5080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2722494324567229e-05,
+      "loss": 0.8655,
+      "step": 5081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719852960023043e-05,
+      "loss": 0.9105,
+      "step": 5082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717211390545065e-05,
+      "loss": 0.9277,
+      "step": 5083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271456961633233e-05,
+      "loss": 0.9022,
+      "step": 5084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711927637583892e-05,
+      "loss": 0.9232,
+      "step": 5085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709285454498812e-05,
+      "loss": 0.8547,
+      "step": 5086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2706643067276177e-05,
+      "loss": 0.8825,
+      "step": 5087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2704000476115079e-05,
+      "loss": 0.9016,
+      "step": 5088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701357681214637e-05,
+      "loss": 0.8761,
+      "step": 5089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2698714682773971e-05,
+      "loss": 0.9003,
+      "step": 5090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2696071480992229e-05,
+      "loss": 0.8976,
+      "step": 5091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2693428076068567e-05,
+      "loss": 0.9161,
+      "step": 5092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.269078446820216e-05,
+      "loss": 0.8976,
+      "step": 5093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2688140657592195e-05,
+      "loss": 0.9247,
+      "step": 5094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2685496644437879e-05,
+      "loss": 0.9198,
+      "step": 5095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2682852428938429e-05,
+      "loss": 0.886,
+      "step": 5096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268020801129308e-05,
+      "loss": 0.9238,
+      "step": 5097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2677563391701083e-05,
+      "loss": 0.8679,
+      "step": 5098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2674918570361706e-05,
+      "loss": 0.9404,
+      "step": 5099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2672273547474225e-05,
+      "loss": 0.9454,
+      "step": 5100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2669628323237937e-05,
+      "loss": 0.8882,
+      "step": 5101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266698289785215e-05,
+      "loss": 0.9223,
+      "step": 5102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2664337271516194e-05,
+      "loss": 0.9136,
+      "step": 5103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661691444429409e-05,
+      "loss": 0.8963,
+      "step": 5104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2659045416791149e-05,
+      "loss": 0.8592,
+      "step": 5105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2656399188800788e-05,
+      "loss": 0.9341,
+      "step": 5106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653752760657709e-05,
+      "loss": 0.9496,
+      "step": 5107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2651106132561317e-05,
+      "loss": 0.9126,
+      "step": 5108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648459304711026e-05,
+      "loss": 0.8837,
+      "step": 5109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264581227730627e-05,
+      "loss": 0.9372,
+      "step": 5110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2643165050546495e-05,
+      "loss": 0.9305,
+      "step": 5111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264051762463116e-05,
+      "loss": 0.6536,
+      "step": 5112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637869999759742e-05,
+      "loss": 0.9419,
+      "step": 5113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2635222176131735e-05,
+      "loss": 0.9163,
+      "step": 5114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2632574153946646e-05,
+      "loss": 0.9576,
+      "step": 5115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262992593340399e-05,
+      "loss": 0.9093,
+      "step": 5116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627277514703313e-05,
+      "loss": 0.9396,
+      "step": 5117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262462889804416e-05,
+      "loss": 0.8921,
+      "step": 5118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26219800836261e-05,
+      "loss": 0.9372,
+      "step": 5119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2619331071648716e-05,
+      "loss": 0.8828,
+      "step": 5120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26166818623116e-05,
+      "loss": 0.9355,
+      "step": 5121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2614032455814362e-05,
+      "loss": 0.9363,
+      "step": 5122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2611382852356632e-05,
+      "loss": 0.8518,
+      "step": 5123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260873305213805e-05,
+      "loss": 0.8943,
+      "step": 5124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260608305535827e-05,
+      "loss": 0.9618,
+      "step": 5125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260343286221696e-05,
+      "loss": 0.8882,
+      "step": 5126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2600782472913811e-05,
+      "loss": 0.9209,
+      "step": 5127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2598131887648516e-05,
+      "loss": 0.9473,
+      "step": 5128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595481106620799e-05,
+      "loss": 0.8932,
+      "step": 5129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259283013003038e-05,
+      "loss": 0.9156,
+      "step": 5130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.259017895807701e-05,
+      "loss": 0.8837,
+      "step": 5131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2587527590960444e-05,
+      "loss": 0.898,
+      "step": 5132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2584876028880455e-05,
+      "loss": 0.8971,
+      "step": 5133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2582224272036834e-05,
+      "loss": 0.8703,
+      "step": 5134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2579572320629381e-05,
+      "loss": 0.9667,
+      "step": 5135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2576920174857917e-05,
+      "loss": 0.9313,
+      "step": 5136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.257426783492227e-05,
+      "loss": 0.9662,
+      "step": 5137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2571615301022288e-05,
+      "loss": 0.9043,
+      "step": 5138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2568962573357837e-05,
+      "loss": 0.9452,
+      "step": 5139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2566309652128789e-05,
+      "loss": 0.9381,
+      "step": 5140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2563656537535036e-05,
+      "loss": 0.9178,
+      "step": 5141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561003229776485e-05,
+      "loss": 0.5615,
+      "step": 5142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2558349729053052e-05,
+      "loss": 0.9372,
+      "step": 5143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255569603556467e-05,
+      "loss": 0.8776,
+      "step": 5144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553042149511295e-05,
+      "loss": 0.9051,
+      "step": 5145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550388071092883e-05,
+      "loss": 0.8658,
+      "step": 5146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547733800509414e-05,
+      "loss": 0.8881,
+      "step": 5147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2545079337960883e-05,
+      "loss": 0.9362,
+      "step": 5148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542424683647294e-05,
+      "loss": 0.9318,
+      "step": 5149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2539769837768672e-05,
+      "loss": 0.9313,
+      "step": 5150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537114800525047e-05,
+      "loss": 0.5175,
+      "step": 5151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2534459572116476e-05,
+      "loss": 0.8967,
+      "step": 5152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531804152743017e-05,
+      "loss": 0.9076,
+      "step": 5153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252914854260475e-05,
+      "loss": 0.902,
+      "step": 5154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252649274190177e-05,
+      "loss": 0.9047,
+      "step": 5155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523836750834186e-05,
+      "loss": 0.9076,
+      "step": 5156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521180569602117e-05,
+      "loss": 0.9323,
+      "step": 5157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518524198405699e-05,
+      "loss": 0.9371,
+      "step": 5158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515867637445088e-05,
+      "loss": 0.9416,
+      "step": 5159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251321088692044e-05,
+      "loss": 0.9154,
+      "step": 5160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510553947031942e-05,
+      "loss": 0.944,
+      "step": 5161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507896817979785e-05,
+      "loss": 0.883,
+      "step": 5162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505239499964179e-05,
+      "loss": 0.9256,
+      "step": 5163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250258199318534e-05,
+      "loss": 0.9736,
+      "step": 5164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249992429784351e-05,
+      "loss": 0.933,
+      "step": 5165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2497266414138935e-05,
+      "loss": 0.9281,
+      "step": 5166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494608342271881e-05,
+      "loss": 0.9329,
+      "step": 5167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249195008244263e-05,
+      "loss": 0.9098,
+      "step": 5168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248929163485147e-05,
+      "loss": 0.9059,
+      "step": 5169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486632999698709e-05,
+      "loss": 0.8676,
+      "step": 5170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2483974177184673e-05,
+      "loss": 0.9564,
+      "step": 5171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2481315167509691e-05,
+      "loss": 0.9369,
+      "step": 5172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247865597087412e-05,
+      "loss": 0.5491,
+      "step": 5173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2475996587478311e-05,
+      "loss": 0.8979,
+      "step": 5174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2473337017522653e-05,
+      "loss": 0.9487,
+      "step": 5175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2470677261207534e-05,
+      "loss": 0.9252,
+      "step": 5176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2468017318733358e-05,
+      "loss": 0.5557,
+      "step": 5177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246535719030055e-05,
+      "loss": 0.9548,
+      "step": 5178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2462696876109534e-05,
+      "loss": 0.9481,
+      "step": 5179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2460036376360768e-05,
+      "loss": 0.9266,
+      "step": 5180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2457375691254707e-05,
+      "loss": 0.9096,
+      "step": 5181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2454714820991828e-05,
+      "loss": 0.9309,
+      "step": 5182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2452053765772623e-05,
+      "loss": 0.8684,
+      "step": 5183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2449392525797597e-05,
+      "loss": 0.8904,
+      "step": 5184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2446731101267262e-05,
+      "loss": 0.8997,
+      "step": 5185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2444069492382151e-05,
+      "loss": 0.8857,
+      "step": 5186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244140769934281e-05,
+      "loss": 0.9213,
+      "step": 5187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24387457223498e-05,
+      "loss": 0.9472,
+      "step": 5188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.243608356160369e-05,
+      "loss": 0.9256,
+      "step": 5189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2433421217305072e-05,
+      "loss": 0.9144,
+      "step": 5190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2430758689654542e-05,
+      "loss": 0.9024,
+      "step": 5191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2428095978852717e-05,
+      "loss": 0.8689,
+      "step": 5192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2425433085100224e-05,
+      "loss": 0.8959,
+      "step": 5193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2422770008597707e-05,
+      "loss": 0.8955,
+      "step": 5194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.242010674954582e-05,
+      "loss": 0.938,
+      "step": 5195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417443308145231e-05,
+      "loss": 0.9064,
+      "step": 5196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414779684596625e-05,
+      "loss": 0.9047,
+      "step": 5197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2412115879100702e-05,
+      "loss": 0.9058,
+      "step": 5198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240945189185817e-05,
+      "loss": 0.9235,
+      "step": 5199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240678772306975e-05,
+      "loss": 0.9083,
+      "step": 5200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2404123372936186e-05,
+      "loss": 0.8711,
+      "step": 5201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401458841658227e-05,
+      "loss": 0.9586,
+      "step": 5202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239879412943664e-05,
+      "loss": 0.9314,
+      "step": 5203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2396129236472201e-05,
+      "loss": 0.944,
+      "step": 5204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2393464162965708e-05,
+      "loss": 0.9764,
+      "step": 5205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2390798909117963e-05,
+      "loss": 0.929,
+      "step": 5206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388133475129784e-05,
+      "loss": 0.9006,
+      "step": 5207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238546786120201e-05,
+      "loss": 0.9284,
+      "step": 5208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382802067535487e-05,
+      "loss": 0.9309,
+      "step": 5209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238013609433107e-05,
+      "loss": 0.8616,
+      "step": 5210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2377469941789639e-05,
+      "loss": 0.9134,
+      "step": 5211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374803610112081e-05,
+      "loss": 0.9193,
+      "step": 5212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2372137099499293e-05,
+      "loss": 0.8663,
+      "step": 5213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2369470410152195e-05,
+      "loss": 0.9395,
+      "step": 5214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2366803542271712e-05,
+      "loss": 0.9409,
+      "step": 5215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2364136496058787e-05,
+      "loss": 0.8886,
+      "step": 5216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2361469271714372e-05,
+      "loss": 0.9318,
+      "step": 5217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2358801869439439e-05,
+      "loss": 0.9468,
+      "step": 5218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356134289434964e-05,
+      "loss": 0.8713,
+      "step": 5219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235346653190195e-05,
+      "loss": 0.8359,
+      "step": 5220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23507985970414e-05,
+      "loss": 0.8771,
+      "step": 5221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348130485054337e-05,
+      "loss": 0.9129,
+      "step": 5222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2345462196141797e-05,
+      "loss": 0.9072,
+      "step": 5223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234279373050483e-05,
+      "loss": 0.8825,
+      "step": 5224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2340125088344497e-05,
+      "loss": 0.9011,
+      "step": 5225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233745626986187e-05,
+      "loss": 0.9356,
+      "step": 5226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233478727525804e-05,
+      "loss": 0.9638,
+      "step": 5227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233211810473411e-05,
+      "loss": 0.9266,
+      "step": 5228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329448758491195e-05,
+      "loss": 0.9178,
+      "step": 5229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232677923673042e-05,
+      "loss": 0.9133,
+      "step": 5230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2324109539652928e-05,
+      "loss": 0.8982,
+      "step": 5231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2321439667459876e-05,
+      "loss": 0.899,
+      "step": 5232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231876962035243e-05,
+      "loss": 0.9339,
+      "step": 5233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231609939853177e-05,
+      "loss": 0.9236,
+      "step": 5234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2313429002199088e-05,
+      "loss": 0.9123,
+      "step": 5235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23107584315556e-05,
+      "loss": 0.9234,
+      "step": 5236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2308087686802521e-05,
+      "loss": 0.9308,
+      "step": 5237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305416768141082e-05,
+      "loss": 0.9172,
+      "step": 5238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2302745675772534e-05,
+      "loss": 0.8967,
+      "step": 5239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300074409898134e-05,
+      "loss": 0.9659,
+      "step": 5240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2297402970719157e-05,
+      "loss": 0.5442,
+      "step": 5241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2294731358436887e-05,
+      "loss": 0.8832,
+      "step": 5242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292059573252625e-05,
+      "loss": 0.898,
+      "step": 5243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2289387615367684e-05,
+      "loss": 0.9085,
+      "step": 5244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2286715484983383e-05,
+      "loss": 0.921,
+      "step": 5245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2284043182301066e-05,
+      "loss": 0.8964,
+      "step": 5246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2281370707522083e-05,
+      "loss": 0.9073,
+      "step": 5247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2278698060847795e-05,
+      "loss": 0.8915,
+      "step": 5248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2276025242479584e-05,
+      "loss": 0.9139,
+      "step": 5249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2273352252618834e-05,
+      "loss": 0.5628,
+      "step": 5250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2270679091466953e-05,
+      "loss": 0.9486,
+      "step": 5251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2268005759225352e-05,
+      "loss": 0.9527,
+      "step": 5252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2265332256095463e-05,
+      "loss": 0.894,
+      "step": 5253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2262658582278726e-05,
+      "loss": 0.9283,
+      "step": 5254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2259984737976597e-05,
+      "loss": 0.8908,
+      "step": 5255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2257310723390541e-05,
+      "loss": 0.8533,
+      "step": 5256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.225463653872204e-05,
+      "loss": 0.9109,
+      "step": 5257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2251962184172586e-05,
+      "loss": 0.9074,
+      "step": 5258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2249287659943682e-05,
+      "loss": 0.8624,
+      "step": 5259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2246612966236851e-05,
+      "loss": 0.9395,
+      "step": 5260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2243938103253618e-05,
+      "loss": 0.8991,
+      "step": 5261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2241263071195535e-05,
+      "loss": 0.9054,
+      "step": 5262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2238587870264152e-05,
+      "loss": 0.9212,
+      "step": 5263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.223591250066104e-05,
+      "loss": 0.8889,
+      "step": 5264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233236962587782e-05,
+      "loss": 0.8948,
+      "step": 5265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2230561256245974e-05,
+      "loss": 0.9093,
+      "step": 5266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222788538183722e-05,
+      "loss": 0.898,
+      "step": 5267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.9137,
+      "step": 5268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222533129625378e-05,
+      "loss": 0.9267,
+      "step": 5269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2219856752225565e-05,
+      "loss": 0.8922,
+      "step": 5270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2217180207565365e-05,
+      "loss": 0.9644,
+      "step": 5271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2214503495846448e-05,
+      "loss": 0.8927,
+      "step": 5272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2211826617270495e-05,
+      "loss": 0.8936,
+      "step": 5273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2209149572039207e-05,
+      "loss": 0.8902,
+      "step": 5274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2206472360354286e-05,
+      "loss": 0.8603,
+      "step": 5275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220379498241746e-05,
+      "loss": 0.9466,
+      "step": 5276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2201117438430456e-05,
+      "loss": 0.939,
+      "step": 5277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2198439728595025e-05,
+      "loss": 0.964,
+      "step": 5278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2195761853112922e-05,
+      "loss": 0.9352,
+      "step": 5279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2193083812185919e-05,
+      "loss": 0.9312,
+      "step": 5280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2190405606015798e-05,
+      "loss": 0.9428,
+      "step": 5281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2187727234804356e-05,
+      "loss": 0.9347,
+      "step": 5282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2185048698753403e-05,
+      "loss": 0.8897,
+      "step": 5283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2182369998064755e-05,
+      "loss": 0.9225,
+      "step": 5284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2179691132940252e-05,
+      "loss": 0.9308,
+      "step": 5285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2177012103581733e-05,
+      "loss": 0.9034,
+      "step": 5286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217433291019106e-05,
+      "loss": 0.9291,
+      "step": 5287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21716535529701e-05,
+      "loss": 0.8709,
+      "step": 5288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2168974032120737e-05,
+      "loss": 0.9015,
+      "step": 5289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2166294347844867e-05,
+      "loss": 0.9177,
+      "step": 5290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163614500344396e-05,
+      "loss": 0.8925,
+      "step": 5291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2160934489821244e-05,
+      "loss": 0.8789,
+      "step": 5292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215825431647734e-05,
+      "loss": 0.4504,
+      "step": 5293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2155573980514632e-05,
+      "loss": 0.8682,
+      "step": 5294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2152893482135075e-05,
+      "loss": 0.942,
+      "step": 5295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2150212821540637e-05,
+      "loss": 0.9217,
+      "step": 5296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21475319989333e-05,
+      "loss": 0.9185,
+      "step": 5297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2144851014515055e-05,
+      "loss": 0.9455,
+      "step": 5298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214216986848791e-05,
+      "loss": 0.8766,
+      "step": 5299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2139488561053879e-05,
+      "loss": 0.9414,
+      "step": 5300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2136807092414992e-05,
+      "loss": 0.9357,
+      "step": 5301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2134125462773293e-05,
+      "loss": 0.9032,
+      "step": 5302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2131443672330835e-05,
+      "loss": 0.8467,
+      "step": 5303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212876172128968e-05,
+      "loss": 0.8694,
+      "step": 5304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2126079609851911e-05,
+      "loss": 0.8382,
+      "step": 5305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2123397338219615e-05,
+      "loss": 0.9309,
+      "step": 5306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120714906594897e-05,
+      "loss": 0.973,
+      "step": 5307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2118032315179871e-05,
+      "loss": 0.9543,
+      "step": 5308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211534956417666e-05,
+      "loss": 0.8918,
+      "step": 5309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112666653787404e-05,
+      "loss": 0.8978,
+      "step": 5310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109983584214253e-05,
+      "loss": 0.8882,
+      "step": 5311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2107300355659368e-05,
+      "loss": 0.9334,
+      "step": 5312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104616968324928e-05,
+      "loss": 0.8876,
+      "step": 5313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2101933422413115e-05,
+      "loss": 0.945,
+      "step": 5314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2099249718126127e-05,
+      "loss": 0.5485,
+      "step": 5315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2096565855666178e-05,
+      "loss": 0.8967,
+      "step": 5316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2093881835235485e-05,
+      "loss": 0.9155,
+      "step": 5317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091197657036285e-05,
+      "loss": 0.8681,
+      "step": 5318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2088513321270823e-05,
+      "loss": 0.8606,
+      "step": 5319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208582882814136e-05,
+      "loss": 0.9033,
+      "step": 5320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083144177850156e-05,
+      "loss": 0.907,
+      "step": 5321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2080459370599502e-05,
+      "loss": 0.9108,
+      "step": 5322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077774406591687e-05,
+      "loss": 0.9057,
+      "step": 5323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2075089286029015e-05,
+      "loss": 0.9422,
+      "step": 5324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072404009113808e-05,
+      "loss": 0.9043,
+      "step": 5325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2069718576048388e-05,
+      "loss": 0.8923,
+      "step": 5326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20670329870351e-05,
+      "loss": 0.8898,
+      "step": 5327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064347242276293e-05,
+      "loss": 0.9163,
+      "step": 5328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2061661341974333e-05,
+      "loss": 0.9018,
+      "step": 5329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2058975286331596e-05,
+      "loss": 0.8942,
+      "step": 5330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2056289075550467e-05,
+      "loss": 0.8404,
+      "step": 5331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2053602709833347e-05,
+      "loss": 0.9263,
+      "step": 5332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050916189382646e-05,
+      "loss": 0.9346,
+      "step": 5333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2048229514400785e-05,
+      "loss": 0.9454,
+      "step": 5334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045542685090199e-05,
+      "loss": 0.9086,
+      "step": 5335
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042855701653333e-05,
+      "loss": 0.9599,
+      "step": 5336
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040168564292644e-05,
+      "loss": 0.9269,
+      "step": 5337
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2037481273210603e-05,
+      "loss": 0.863,
+      "step": 5338
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203479382860969e-05,
+      "loss": 0.9279,
+      "step": 5339
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032106230692394e-05,
+      "loss": 0.881,
+      "step": 5340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2029418479661222e-05,
+      "loss": 0.9244,
+      "step": 5341
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026730575718684e-05,
+      "loss": 0.873,
+      "step": 5342
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202404251906731e-05,
+      "loss": 0.8938,
+      "step": 5343
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2021354309909637e-05,
+      "loss": 0.5037,
+      "step": 5344
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018665948448215e-05,
+      "loss": 0.9031,
+      "step": 5345
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2015977434885608e-05,
+      "loss": 0.9059,
+      "step": 5346
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2013288769424383e-05,
+      "loss": 0.9688,
+      "step": 5347
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010599952267128e-05,
+      "loss": 0.9463,
+      "step": 5348
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007910983616435e-05,
+      "loss": 0.9678,
+      "step": 5349
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2005221863674914e-05,
+      "loss": 0.9109,
+      "step": 5350
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002532592645182e-05,
+      "loss": 0.5038,
+      "step": 5351
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1999843170729866e-05,
+      "loss": 0.9741,
+      "step": 5352
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199715359813161e-05,
+      "loss": 0.9224,
+      "step": 5353
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994463875053065e-05,
+      "loss": 0.9413,
+      "step": 5354
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1991774001696896e-05,
+      "loss": 0.8826,
+      "step": 5355
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1989083978265776e-05,
+      "loss": 0.918,
+      "step": 5356
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1986393804962393e-05,
+      "loss": 0.8875,
+      "step": 5357
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1983703481989443e-05,
+      "loss": 0.8489,
+      "step": 5358
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981013009549636e-05,
+      "loss": 0.9061,
+      "step": 5359
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1978322387845689e-05,
+      "loss": 0.9198,
+      "step": 5360
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1975631617080339e-05,
+      "loss": 0.8981,
+      "step": 5361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1972940697456324e-05,
+      "loss": 0.9404,
+      "step": 5362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1970249629176399e-05,
+      "loss": 0.6395,
+      "step": 5363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1967558412443328e-05,
+      "loss": 0.9271,
+      "step": 5364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1964867047459887e-05,
+      "loss": 0.9181,
+      "step": 5365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1962175534428865e-05,
+      "loss": 0.9046,
+      "step": 5366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1959483873553059e-05,
+      "loss": 0.8504,
+      "step": 5367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1956792065035281e-05,
+      "loss": 0.9376,
+      "step": 5368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.195410010907835e-05,
+      "loss": 0.9137,
+      "step": 5369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1951408005885098e-05,
+      "loss": 0.8844,
+      "step": 5370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1948715755658364e-05,
+      "loss": 0.9269,
+      "step": 5371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.194602335860101e-05,
+      "loss": 0.9312,
+      "step": 5372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1943330814915897e-05,
+      "loss": 0.8877,
+      "step": 5373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.19406381248059e-05,
+      "loss": 0.9604,
+      "step": 5374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1937945288473905e-05,
+      "loss": 0.5195,
+      "step": 5375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1935252306122812e-05,
+      "loss": 0.9541,
+      "step": 5376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1932559177955533e-05,
+      "loss": 0.9192,
+      "step": 5377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1929865904174984e-05,
+      "loss": 0.8917,
+      "step": 5378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.19271724849841e-05,
+      "loss": 0.9046,
+      "step": 5379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1924478920585818e-05,
+      "loss": 0.8936,
+      "step": 5380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921785211183095e-05,
+      "loss": 0.8755,
+      "step": 5381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1919091356978894e-05,
+      "loss": 0.9193,
+      "step": 5382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1916397358176191e-05,
+      "loss": 0.9041,
+      "step": 5383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913703214977969e-05,
+      "loss": 0.9376,
+      "step": 5384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1911008927587224e-05,
+      "loss": 0.9045,
+      "step": 5385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1908314496206967e-05,
+      "loss": 0.8992,
+      "step": 5386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905619921040215e-05,
+      "loss": 0.8507,
+      "step": 5387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902925202289997e-05,
+      "loss": 0.9168,
+      "step": 5388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1900230340159352e-05,
+      "loss": 0.9319,
+      "step": 5389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1897535334851331e-05,
+      "loss": 0.8796,
+      "step": 5390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1894840186569e-05,
+      "loss": 0.916,
+      "step": 5391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892144895515422e-05,
+      "loss": 0.9541,
+      "step": 5392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1889449461893692e-05,
+      "loss": 0.9098,
+      "step": 5393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1886753885906895e-05,
+      "loss": 0.9006,
+      "step": 5394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884058167758137e-05,
+      "loss": 0.8781,
+      "step": 5395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1881362307650534e-05,
+      "loss": 0.5723,
+      "step": 5396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1878666305787214e-05,
+      "loss": 0.9107,
+      "step": 5397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1875970162371312e-05,
+      "loss": 0.8977,
+      "step": 5398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1873273877605978e-05,
+      "loss": 0.8704,
+      "step": 5399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1870577451694363e-05,
+      "loss": 0.8937,
+      "step": 5400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1867880884839642e-05,
+      "loss": 0.9497,
+      "step": 5401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1865184177244992e-05,
+      "loss": 0.8821,
+      "step": 5402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1862487329113606e-05,
+      "loss": 0.8877,
+      "step": 5403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185979034064868e-05,
+      "loss": 0.9047,
+      "step": 5404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857093212053427e-05,
+      "loss": 0.9079,
+      "step": 5405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185439594353107e-05,
+      "loss": 0.8903,
+      "step": 5406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851698535284834e-05,
+      "loss": 0.5818,
+      "step": 5407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184900098751797e-05,
+      "loss": 0.8897,
+      "step": 5408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184630330043373e-05,
+      "loss": 0.9482,
+      "step": 5409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1843605474235373e-05,
+      "loss": 0.9232,
+      "step": 5410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1840907509126178e-05,
+      "loss": 0.9623,
+      "step": 5411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838209405309427e-05,
+      "loss": 0.9371,
+      "step": 5412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1835511162988413e-05,
+      "loss": 0.9184,
+      "step": 5413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1832812782366448e-05,
+      "loss": 0.8976,
+      "step": 5414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1830114263646844e-05,
+      "loss": 0.9563,
+      "step": 5415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827415607032923e-05,
+      "loss": 0.9505,
+      "step": 5416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1824716812728029e-05,
+      "loss": 0.892,
+      "step": 5417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1822017880935507e-05,
+      "loss": 0.9305,
+      "step": 5418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1819318811858711e-05,
+      "loss": 0.9055,
+      "step": 5419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816619605701012e-05,
+      "loss": 0.8887,
+      "step": 5420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1813920262665788e-05,
+      "loss": 0.8985,
+      "step": 5421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1811220782956427e-05,
+      "loss": 0.9263,
+      "step": 5422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1808521166776328e-05,
+      "loss": 0.9543,
+      "step": 5423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805821414328897e-05,
+      "loss": 0.9344,
+      "step": 5424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180312152581756e-05,
+      "loss": 0.9321,
+      "step": 5425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1800421501445743e-05,
+      "loss": 0.9061,
+      "step": 5426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797721341416882e-05,
+      "loss": 0.9213,
+      "step": 5427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179502104593443e-05,
+      "loss": 0.8951,
+      "step": 5428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1792320615201846e-05,
+      "loss": 0.9233,
+      "step": 5429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789620049422603e-05,
+      "loss": 0.9415,
+      "step": 5430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786919348800181e-05,
+      "loss": 0.889,
+      "step": 5431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178421851353807e-05,
+      "loss": 0.8766,
+      "step": 5432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178151754383977e-05,
+      "loss": 0.8721,
+      "step": 5433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778816439908794e-05,
+      "loss": 0.8693,
+      "step": 5434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1776115201948658e-05,
+      "loss": 0.9261,
+      "step": 5435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1773413830162902e-05,
+      "loss": 0.585,
+      "step": 5436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770712324755063e-05,
+      "loss": 0.9271,
+      "step": 5437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770712324755063e-05,
+      "loss": 0.9127,
+      "step": 5438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1768010685928686e-05,
+      "loss": 0.9346,
+      "step": 5439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765308913887341e-05,
+      "loss": 0.9336,
+      "step": 5440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1762607008834597e-05,
+      "loss": 0.8971,
+      "step": 5441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759904970974034e-05,
+      "loss": 0.9313,
+      "step": 5442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1757202800509245e-05,
+      "loss": 0.8944,
+      "step": 5443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1754500497643831e-05,
+      "loss": 0.9406,
+      "step": 5444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1751798062581402e-05,
+      "loss": 0.8822,
+      "step": 5445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749095495525584e-05,
+      "loss": 0.9334,
+      "step": 5446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1746392796680004e-05,
+      "loss": 0.9439,
+      "step": 5447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1743689966248304e-05,
+      "loss": 0.9208,
+      "step": 5448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1740987004434137e-05,
+      "loss": 0.89,
+      "step": 5449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1738283911441159e-05,
+      "loss": 0.9147,
+      "step": 5450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1735580687473047e-05,
+      "loss": 0.8751,
+      "step": 5451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1732877332733479e-05,
+      "loss": 0.8991,
+      "step": 5452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730173847426148e-05,
+      "loss": 0.9128,
+      "step": 5453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1727470231754749e-05,
+      "loss": 0.8616,
+      "step": 5454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1724766485922998e-05,
+      "loss": 0.9183,
+      "step": 5455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722062610134615e-05,
+      "loss": 0.9116,
+      "step": 5456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1719358604593326e-05,
+      "loss": 0.8595,
+      "step": 5457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1716654469502875e-05,
+      "loss": 0.8691,
+      "step": 5458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1713950205067008e-05,
+      "loss": 0.894,
+      "step": 5459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1711245811489485e-05,
+      "loss": 0.923,
+      "step": 5460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1708541288974074e-05,
+      "loss": 0.9256,
+      "step": 5461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1705836637724558e-05,
+      "loss": 0.9251,
+      "step": 5462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170313185794472e-05,
+      "loss": 0.9649,
+      "step": 5463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700426949838364e-05,
+      "loss": 0.9078,
+      "step": 5464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1697721913609295e-05,
+      "loss": 0.9054,
+      "step": 5465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1695016749461328e-05,
+      "loss": 0.9263,
+      "step": 5466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1692311457598291e-05,
+      "loss": 0.9403,
+      "step": 5467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1689606038224024e-05,
+      "loss": 0.8143,
+      "step": 5468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168690049154237e-05,
+      "loss": 0.9581,
+      "step": 5469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1684194817757184e-05,
+      "loss": 0.9336,
+      "step": 5470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1681489017072333e-05,
+      "loss": 0.8871,
+      "step": 5471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1678783089691695e-05,
+      "loss": 0.9117,
+      "step": 5472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1676077035819148e-05,
+      "loss": 0.9037,
+      "step": 5473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1673370855658592e-05,
+      "loss": 0.8881,
+      "step": 5474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670664549413928e-05,
+      "loss": 0.9401,
+      "step": 5475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1667958117289068e-05,
+      "loss": 0.8843,
+      "step": 5476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1665251559487938e-05,
+      "loss": 0.8661,
+      "step": 5477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1662544876214468e-05,
+      "loss": 0.8844,
+      "step": 5478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16598380676726e-05,
+      "loss": 0.9212,
+      "step": 5479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1657131134066281e-05,
+      "loss": 0.9053,
+      "step": 5480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1654424075599478e-05,
+      "loss": 0.93,
+      "step": 5481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651716892476154e-05,
+      "loss": 0.9128,
+      "step": 5482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1649009584900293e-05,
+      "loss": 0.8807,
+      "step": 5483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646302153075882e-05,
+      "loss": 0.9128,
+      "step": 5484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164359459720692e-05,
+      "loss": 0.9036,
+      "step": 5485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640886917497412e-05,
+      "loss": 0.8694,
+      "step": 5486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1638179114151378e-05,
+      "loss": 0.8836,
+      "step": 5487
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.163547118737284e-05,
+      "loss": 0.9203,
+      "step": 5488
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1632763137365837e-05,
+      "loss": 0.9259,
+      "step": 5489
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.163005496433441e-05,
+      "loss": 0.9567,
+      "step": 5490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1627346668482614e-05,
+      "loss": 0.9046,
+      "step": 5491
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.162463825001451e-05,
+      "loss": 0.915,
+      "step": 5492
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1621929709134177e-05,
+      "loss": 0.9237,
+      "step": 5493
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1619221046045688e-05,
+      "loss": 0.8647,
+      "step": 5494
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1616512260953139e-05,
+      "loss": 0.9058,
+      "step": 5495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1613803354060626e-05,
+      "loss": 0.9382,
+      "step": 5496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1611094325572263e-05,
+      "loss": 0.9157,
+      "step": 5497
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1608385175692165e-05,
+      "loss": 0.9109,
+      "step": 5498
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160567590462446e-05,
+      "loss": 0.9433,
+      "step": 5499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1602966512573286e-05,
+      "loss": 0.8673,
+      "step": 5500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1600256999742785e-05,
+      "loss": 0.8081,
+      "step": 5501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597547366337114e-05,
+      "loss": 0.9463,
+      "step": 5502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1594837612560437e-05,
+      "loss": 0.8739,
+      "step": 5503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1592127738616925e-05,
+      "loss": 0.8576,
+      "step": 5504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1589417744710758e-05,
+      "loss": 0.8583,
+      "step": 5505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586707631046135e-05,
+      "loss": 0.9604,
+      "step": 5506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158399739782725e-05,
+      "loss": 0.9026,
+      "step": 5507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1581287045258312e-05,
+      "loss": 0.9093,
+      "step": 5508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1578576573543541e-05,
+      "loss": 0.9115,
+      "step": 5509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1575865982887164e-05,
+      "loss": 0.9093,
+      "step": 5510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573155273493415e-05,
+      "loss": 0.8985,
+      "step": 5511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1570444445566538e-05,
+      "loss": 0.9407,
+      "step": 5512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156773349931079e-05,
+      "loss": 0.8844,
+      "step": 5513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1565022434930433e-05,
+      "loss": 0.5088,
+      "step": 5514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1562311252629736e-05,
+      "loss": 0.5697,
+      "step": 5515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1559599952612981e-05,
+      "loss": 0.9193,
+      "step": 5516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1556888535084458e-05,
+      "loss": 0.9207,
+      "step": 5517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1554177000248466e-05,
+      "loss": 0.8638,
+      "step": 5518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155146534830931e-05,
+      "loss": 0.873,
+      "step": 5519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1548753579471306e-05,
+      "loss": 0.9465,
+      "step": 5520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546041693938784e-05,
+      "loss": 0.9108,
+      "step": 5521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543329691916067e-05,
+      "loss": 0.9475,
+      "step": 5522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1540617573607506e-05,
+      "loss": 0.9473,
+      "step": 5523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1537905339217448e-05,
+      "loss": 0.9387,
+      "step": 5524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1535192988950254e-05,
+      "loss": 0.9171,
+      "step": 5525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532480523010292e-05,
+      "loss": 0.8907,
+      "step": 5526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529767941601937e-05,
+      "loss": 0.9231,
+      "step": 5527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527055244929579e-05,
+      "loss": 0.903,
+      "step": 5528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1524342433197614e-05,
+      "loss": 0.885,
+      "step": 5529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1521629506610439e-05,
+      "loss": 0.9206,
+      "step": 5530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1518916465372474e-05,
+      "loss": 0.8859,
+      "step": 5531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151620330968813e-05,
+      "loss": 0.9247,
+      "step": 5532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513490039761843e-05,
+      "loss": 0.9003,
+      "step": 5533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510776655798045e-05,
+      "loss": 0.9238,
+      "step": 5534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.150806315800119e-05,
+      "loss": 0.9184,
+      "step": 5535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1505349546575728e-05,
+      "loss": 0.8966,
+      "step": 5536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502635821726125e-05,
+      "loss": 0.9149,
+      "step": 5537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1499921983656848e-05,
+      "loss": 0.942,
+      "step": 5538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497208032572385e-05,
+      "loss": 0.9052,
+      "step": 5539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1494493968677216e-05,
+      "loss": 0.9382,
+      "step": 5540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1491779792175848e-05,
+      "loss": 0.94,
+      "step": 5541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1489065503272785e-05,
+      "loss": 0.9596,
+      "step": 5542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486351102172534e-05,
+      "loss": 0.847,
+      "step": 5543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1483636589079627e-05,
+      "loss": 0.8832,
+      "step": 5544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148092196419859e-05,
+      "loss": 0.9098,
+      "step": 5545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1478207227733966e-05,
+      "loss": 0.9315,
+      "step": 5546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1475492379890301e-05,
+      "loss": 0.9079,
+      "step": 5547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1472777420872154e-05,
+      "loss": 0.911,
+      "step": 5548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1470062350884086e-05,
+      "loss": 0.9476,
+      "step": 5549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1467347170130674e-05,
+      "loss": 0.8542,
+      "step": 5550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1464631878816502e-05,
+      "loss": 0.9237,
+      "step": 5551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1461916477146156e-05,
+      "loss": 0.9075,
+      "step": 5552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459200965324233e-05,
+      "loss": 0.9024,
+      "step": 5553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1456485343555344e-05,
+      "loss": 0.9154,
+      "step": 5554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14537696120441e-05,
+      "loss": 0.5884,
+      "step": 5555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1451053770995131e-05,
+      "loss": 0.9428,
+      "step": 5556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1448337820613061e-05,
+      "loss": 0.8925,
+      "step": 5557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445621761102532e-05,
+      "loss": 0.9123,
+      "step": 5558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1442905592668191e-05,
+      "loss": 0.9331,
+      "step": 5559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1440189315514698e-05,
+      "loss": 0.9468,
+      "step": 5560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1437472929846715e-05,
+      "loss": 0.9059,
+      "step": 5561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434756435868917e-05,
+      "loss": 0.864,
+      "step": 5562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432039833785979e-05,
+      "loss": 0.9173,
+      "step": 5563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429323123802593e-05,
+      "loss": 0.8709,
+      "step": 5564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1426606306123458e-05,
+      "loss": 0.9246,
+      "step": 5565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423889380953277e-05,
+      "loss": 0.9352,
+      "step": 5566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1421172348496763e-05,
+      "loss": 0.9208,
+      "step": 5567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1418455208958636e-05,
+      "loss": 0.9266,
+      "step": 5568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141573796254363e-05,
+      "loss": 0.921,
+      "step": 5569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413020609456477e-05,
+      "loss": 0.9669,
+      "step": 5570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410303149901924e-05,
+      "loss": 0.8927,
+      "step": 5571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1407585584084726e-05,
+      "loss": 0.8842,
+      "step": 5572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404867912209647e-05,
+      "loss": 0.9652,
+      "step": 5573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402150134481453e-05,
+      "loss": 0.8735,
+      "step": 5574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399432251104918e-05,
+      "loss": 0.8973,
+      "step": 5575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1396714262284833e-05,
+      "loss": 0.9394,
+      "step": 5576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1393996168225989e-05,
+      "loss": 0.9078,
+      "step": 5577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.139127796913319e-05,
+      "loss": 0.9136,
+      "step": 5578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1388559665211241e-05,
+      "loss": 0.9591,
+      "step": 5579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138584125666496e-05,
+      "loss": 0.9283,
+      "step": 5580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1383122743699173e-05,
+      "loss": 0.8984,
+      "step": 5581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1380404126518714e-05,
+      "loss": 0.94,
+      "step": 5582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1377685405328423e-05,
+      "loss": 0.8714,
+      "step": 5583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1374966580333147e-05,
+      "loss": 0.8865,
+      "step": 5584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1372247651737742e-05,
+      "loss": 0.9236,
+      "step": 5585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1369528619747074e-05,
+      "loss": 0.8982,
+      "step": 5586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1366809484566015e-05,
+      "loss": 0.9387,
+      "step": 5587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1364090246399441e-05,
+      "loss": 0.8864,
+      "step": 5588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1361370905452246e-05,
+      "loss": 0.8874,
+      "step": 5589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1358651461929319e-05,
+      "loss": 0.8493,
+      "step": 5590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135593191603557e-05,
+      "loss": 0.9266,
+      "step": 5591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1353212267975901e-05,
+      "loss": 0.9311,
+      "step": 5592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1350492517955234e-05,
+      "loss": 0.9356,
+      "step": 5593
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1347772666178501e-05,
+      "loss": 0.9227,
+      "step": 5594
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1345052712850626e-05,
+      "loss": 0.9291,
+      "step": 5595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1342332658176556e-05,
+      "loss": 0.9316,
+      "step": 5596
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133961250236124e-05,
+      "loss": 0.8629,
+      "step": 5597
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133689224560963e-05,
+      "loss": 0.9355,
+      "step": 5598
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1334171888126698e-05,
+      "loss": 0.8842,
+      "step": 5599
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1331451430117406e-05,
+      "loss": 0.9341,
+      "step": 5600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1328730871786742e-05,
+      "loss": 0.8994,
+      "step": 5601
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1326010213339688e-05,
+      "loss": 0.9477,
+      "step": 5602
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1323289454981243e-05,
+      "loss": 0.9291,
+      "step": 5603
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320568596916405e-05,
+      "loss": 0.9312,
+      "step": 5604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1317847639350186e-05,
+      "loss": 0.8705,
+      "step": 5605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1315126582487599e-05,
+      "loss": 0.9169,
+      "step": 5606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1312405426533672e-05,
+      "loss": 0.8726,
+      "step": 5607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1309684171693435e-05,
+      "loss": 0.888,
+      "step": 5608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1306962818171932e-05,
+      "loss": 0.8654,
+      "step": 5609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1304241366174205e-05,
+      "loss": 0.9369,
+      "step": 5610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1301519815905309e-05,
+      "loss": 0.9085,
+      "step": 5611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1298798167570306e-05,
+      "loss": 0.9131,
+      "step": 5612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1296076421374264e-05,
+      "loss": 0.9089,
+      "step": 5613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1293354577522264e-05,
+      "loss": 0.874,
+      "step": 5614
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1290632636219385e-05,
+      "loss": 0.9544,
+      "step": 5615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287910597670722e-05,
+      "loss": 0.8974,
+      "step": 5616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128518846208137e-05,
+      "loss": 0.9571,
+      "step": 5617
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1282466229656435e-05,
+      "loss": 0.8981,
+      "step": 5618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1279743900601033e-05,
+      "loss": 0.9289,
+      "step": 5619
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127702147512028e-05,
+      "loss": 0.8478,
+      "step": 5620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1274298953419307e-05,
+      "loss": 0.8601,
+      "step": 5621
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127157633570325e-05,
+      "loss": 0.8695,
+      "step": 5622
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1268853622177248e-05,
+      "loss": 0.9417,
+      "step": 5623
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266130813046449e-05,
+      "loss": 0.9495,
+      "step": 5624
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263407908516018e-05,
+      "loss": 0.9193,
+      "step": 5625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260684908791109e-05,
+      "loss": 0.9337,
+      "step": 5626
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257961814076897e-05,
+      "loss": 0.8835,
+      "step": 5627
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1255238624578559e-05,
+      "loss": 0.9352,
+      "step": 5628
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252515340501282e-05,
+      "loss": 0.9391,
+      "step": 5629
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1249791962050256e-05,
+      "loss": 0.883,
+      "step": 5630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247068489430682e-05,
+      "loss": 0.8757,
+      "step": 5631
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244344922847765e-05,
+      "loss": 0.9054,
+      "step": 5632
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1241621262506722e-05,
+      "loss": 0.9155,
+      "step": 5633
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123889750861277e-05,
+      "loss": 0.9231,
+      "step": 5634
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236173661371139e-05,
+      "loss": 0.8925,
+      "step": 5635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1233449720987063e-05,
+      "loss": 0.9234,
+      "step": 5636
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1230725687665783e-05,
+      "loss": 0.8831,
+      "step": 5637
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228001561612547e-05,
+      "loss": 0.8794,
+      "step": 5638
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1225277343032614e-05,
+      "loss": 0.5844,
+      "step": 5639
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222553032131244e-05,
+      "loss": 0.9302,
+      "step": 5640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219828629113707e-05,
+      "loss": 0.8958,
+      "step": 5641
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1217104134185281e-05,
+      "loss": 0.9195,
+      "step": 5642
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121437954755125e-05,
+      "loss": 0.8831,
+      "step": 5643
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1211654869416901e-05,
+      "loss": 0.9019,
+      "step": 5644
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1208930099987536e-05,
+      "loss": 0.9069,
+      "step": 5645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1206205239468457e-05,
+      "loss": 0.9372,
+      "step": 5646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1203480288064974e-05,
+      "loss": 0.8675,
+      "step": 5647
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1200755245982405e-05,
+      "loss": 0.8787,
+      "step": 5648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1198030113426074e-05,
+      "loss": 0.9006,
+      "step": 5649
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195304890601317e-05,
+      "loss": 0.8926,
+      "step": 5650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.119257957771347e-05,
+      "loss": 0.971,
+      "step": 5651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1189854174967876e-05,
+      "loss": 0.8995,
+      "step": 5652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1187128682569888e-05,
+      "loss": 0.9143,
+      "step": 5653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1184403100724868e-05,
+      "loss": 0.8755,
+      "step": 5654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1181677429638176e-05,
+      "loss": 0.9122,
+      "step": 5655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117895166951519e-05,
+      "loss": 0.9232,
+      "step": 5656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1176225820561284e-05,
+      "loss": 0.8747,
+      "step": 5657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173499882981847e-05,
+      "loss": 0.9661,
+      "step": 5658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170773856982268e-05,
+      "loss": 0.9184,
+      "step": 5659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1168047742767949e-05,
+      "loss": 0.8878,
+      "step": 5660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1165321540544294e-05,
+      "loss": 0.9003,
+      "step": 5661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162595250516715e-05,
+      "loss": 0.8834,
+      "step": 5662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115986887289063e-05,
+      "loss": 0.9332,
+      "step": 5663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157142407871469e-05,
+      "loss": 0.9338,
+      "step": 5664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1154415855664657e-05,
+      "loss": 0.8968,
+      "step": 5665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151689216475641e-05,
+      "loss": 0.9053,
+      "step": 5666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114896249050986e-05,
+      "loss": 0.9117,
+      "step": 5667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1146235677972765e-05,
+      "loss": 0.9067,
+      "step": 5668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143508779069817e-05,
+      "loss": 0.9098,
+      "step": 5669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140781794006479e-05,
+      "loss": 0.8462,
+      "step": 5670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138054722988223e-05,
+      "loss": 0.5792,
+      "step": 5671
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1135327566220528e-05,
+      "loss": 0.925,
+      "step": 5672
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1132600323908876e-05,
+      "loss": 0.9179,
+      "step": 5673
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1129872996258757e-05,
+      "loss": 0.8855,
+      "step": 5674
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127145583475671e-05,
+      "loss": 0.9066,
+      "step": 5675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112441808576512e-05,
+      "loss": 0.8996,
+      "step": 5676
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1121690503332613e-05,
+      "loss": 0.919,
+      "step": 5677
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1118962836383667e-05,
+      "loss": 0.8757,
+      "step": 5678
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1116235085123804e-05,
+      "loss": 0.916,
+      "step": 5679
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113507249758553e-05,
+      "loss": 0.8893,
+      "step": 5680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1110779330493448e-05,
+      "loss": 0.9251,
+      "step": 5681
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108051327534035e-05,
+      "loss": 0.8713,
+      "step": 5682
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1105323241085856e-05,
+      "loss": 0.9346,
+      "step": 5683
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102595071354471e-05,
+      "loss": 0.8693,
+      "step": 5684
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1099866818545437e-05,
+      "loss": 0.8832,
+      "step": 5685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109713848286432e-05,
+      "loss": 0.8884,
+      "step": 5686
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1094410064516694e-05,
+      "loss": 0.8871,
+      "step": 5687
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091681563708142e-05,
+      "loss": 0.8868,
+      "step": 5688
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088952980644242e-05,
+      "loss": 0.894,
+      "step": 5689
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086224315530589e-05,
+      "loss": 0.8696,
+      "step": 5690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083495568572781e-05,
+      "loss": 0.8833,
+      "step": 5691
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1080766739976424e-05,
+      "loss": 0.9301,
+      "step": 5692
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1078037829947123e-05,
+      "loss": 0.9415,
+      "step": 5693
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10753088386905e-05,
+      "loss": 0.924,
+      "step": 5694
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1072579766412172e-05,
+      "loss": 0.9366,
+      "step": 5695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1069850613317772e-05,
+      "loss": 0.9332,
+      "step": 5696
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1067121379612933e-05,
+      "loss": 0.8877,
+      "step": 5697
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1064392065503294e-05,
+      "loss": 0.918,
+      "step": 5698
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1061662671194503e-05,
+      "loss": 0.9269,
+      "step": 5699
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058933196892214e-05,
+      "loss": 0.8681,
+      "step": 5700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1056203642802081e-05,
+      "loss": 0.9208,
+      "step": 5701
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1053474009129774e-05,
+      "loss": 0.9479,
+      "step": 5702
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1050744296080962e-05,
+      "loss": 0.8861,
+      "step": 5703
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1048014503861321e-05,
+      "loss": 0.9127,
+      "step": 5704
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 0.885,
+      "step": 5705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042554682732292e-05,
+      "loss": 0.887,
+      "step": 5706
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1039824654234286e-05,
+      "loss": 0.9695,
+      "step": 5707
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1037094547388217e-05,
+      "loss": 0.8849,
+      "step": 5708
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1034364362399794e-05,
+      "loss": 0.8796,
+      "step": 5709
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1031634099474727e-05,
+      "loss": 0.8677,
+      "step": 5710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1028903758818733e-05,
+      "loss": 0.9098,
+      "step": 5711
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1026173340637541e-05,
+      "loss": 0.9256,
+      "step": 5712
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023442845136874e-05,
+      "loss": 0.9042,
+      "step": 5713
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1020712272522473e-05,
+      "loss": 0.9136,
+      "step": 5714
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1017981623000078e-05,
+      "loss": 0.9134,
+      "step": 5715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1015250896775436e-05,
+      "loss": 0.9277,
+      "step": 5716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1012520094054298e-05,
+      "loss": 0.92,
+      "step": 5717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1009789215042425e-05,
+      "loss": 0.892,
+      "step": 5718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007058259945584e-05,
+      "loss": 0.601,
+      "step": 5719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1004327228969543e-05,
+      "loss": 0.5258,
+      "step": 5720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1001596122320076e-05,
+      "loss": 0.9228,
+      "step": 5721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0998864940202967e-05,
+      "loss": 0.878,
+      "step": 5722
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0996133682824004e-05,
+      "loss": 0.9366,
+      "step": 5723
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0993402350388978e-05,
+      "loss": 0.8914,
+      "step": 5724
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0990670943103688e-05,
+      "loss": 0.9101,
+      "step": 5725
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.098793946117394e-05,
+      "loss": 0.8947,
+      "step": 5726
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0985207904805545e-05,
+      "loss": 0.9122,
+      "step": 5727
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0982476274204314e-05,
+      "loss": 0.9065,
+      "step": 5728
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0979744569576072e-05,
+      "loss": 0.9111,
+      "step": 5729
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0977012791126646e-05,
+      "loss": 0.9024,
+      "step": 5730
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0974280939061867e-05,
+      "loss": 0.9309,
+      "step": 5731
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0971549013587573e-05,
+      "loss": 0.9133,
+      "step": 5732
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0968817014909605e-05,
+      "loss": 0.9023,
+      "step": 5733
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0966084943233818e-05,
+      "loss": 0.9398,
+      "step": 5734
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0963352798766058e-05,
+      "loss": 0.8569,
+      "step": 5735
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0960620581712194e-05,
+      "loss": 0.9073,
+      "step": 5736
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0957888292278084e-05,
+      "loss": 0.8468,
+      "step": 5737
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0955155930669604e-05,
+      "loss": 0.9067,
+      "step": 5738
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0952423497092629e-05,
+      "loss": 0.8906,
+      "step": 5739
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0949690991753036e-05,
+      "loss": 0.9067,
+      "step": 5740
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0946958414856721e-05,
+      "loss": 0.8823,
+      "step": 5741
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0944225766609569e-05,
+      "loss": 0.8795,
+      "step": 5742
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0941493047217482e-05,
+      "loss": 0.8813,
+      "step": 5743
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093876025688636e-05,
+      "loss": 0.8914,
+      "step": 5744
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0936027395822114e-05,
+      "loss": 0.8923,
+      "step": 5745
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0933294464230657e-05,
+      "loss": 0.9114,
+      "step": 5746
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0930561462317908e-05,
+      "loss": 0.8784,
+      "step": 5747
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0927828390289794e-05,
+      "loss": 0.5446,
+      "step": 5748
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092509524835224e-05,
+      "loss": 0.938,
+      "step": 5749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0922362036711186e-05,
+      "loss": 0.9046,
+      "step": 5750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091962875557257e-05,
+      "loss": 0.8567,
+      "step": 5751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0916895405142339e-05,
+      "loss": 0.8881,
+      "step": 5752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0914161985626441e-05,
+      "loss": 0.9055,
+      "step": 5753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0911428497230834e-05,
+      "loss": 0.9042,
+      "step": 5754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908694940161477e-05,
+      "loss": 0.9366,
+      "step": 5755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090596131462434e-05,
+      "loss": 0.8879,
+      "step": 5756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903227620825392e-05,
+      "loss": 0.873,
+      "step": 5757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090049385897061e-05,
+      "loss": 0.5653,
+      "step": 5758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897760029265977e-05,
+      "loss": 0.8952,
+      "step": 5759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0895026131917478e-05,
+      "loss": 0.8934,
+      "step": 5760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892292167131107e-05,
+      "loss": 0.9042,
+      "step": 5761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0889558135112858e-05,
+      "loss": 0.9444,
+      "step": 5762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886824036068738e-05,
+      "loss": 0.9329,
+      "step": 5763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0884089870204751e-05,
+      "loss": 0.8783,
+      "step": 5764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881355637726907e-05,
+      "loss": 0.8392,
+      "step": 5765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087862133884123e-05,
+      "loss": 0.9157,
+      "step": 5766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0875886973753735e-05,
+      "loss": 0.8772,
+      "step": 5767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873152542670453e-05,
+      "loss": 0.9109,
+      "step": 5768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870418045797416e-05,
+      "loss": 0.9192,
+      "step": 5769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086768348334066e-05,
+      "loss": 0.876,
+      "step": 5770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0864948855506227e-05,
+      "loss": 0.8546,
+      "step": 5771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862214162500172e-05,
+      "loss": 0.877,
+      "step": 5772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0859479404528532e-05,
+      "loss": 0.8697,
+      "step": 5773
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0856744581797374e-05,
+      "loss": 0.9594,
+      "step": 5774
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854009694512759e-05,
+      "loss": 0.8679,
+      "step": 5775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085127474288075e-05,
+      "loss": 0.9329,
+      "step": 5776
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0848539727107422e-05,
+      "loss": 0.5338,
+      "step": 5777
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845804647398848e-05,
+      "loss": 0.9014,
+      "step": 5778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0843069503961112e-05,
+      "loss": 0.9186,
+      "step": 5779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0840334297000297e-05,
+      "loss": 0.8777,
+      "step": 5780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0837599026722499e-05,
+      "loss": 0.8982,
+      "step": 5781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0834863693333805e-05,
+      "loss": 0.9445,
+      "step": 5782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0832128297040323e-05,
+      "loss": 0.9318,
+      "step": 5783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0829392838048157e-05,
+      "loss": 0.8511,
+      "step": 5784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0826657316563412e-05,
+      "loss": 0.8779,
+      "step": 5785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0823921732792205e-05,
+      "loss": 0.9081,
+      "step": 5786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0821186086940653e-05,
+      "loss": 0.4637,
+      "step": 5787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818450379214887e-05,
+      "loss": 0.894,
+      "step": 5788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0815714609821027e-05,
+      "loss": 0.9254,
+      "step": 5789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0812978778965208e-05,
+      "loss": 0.922,
+      "step": 5790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0810242886853572e-05,
+      "loss": 0.8597,
+      "step": 5791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0807506933692255e-05,
+      "loss": 0.8498,
+      "step": 5792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0804770919687413e-05,
+      "loss": 0.8917,
+      "step": 5793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0802034845045189e-05,
+      "loss": 0.8637,
+      "step": 5794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079929870997174e-05,
+      "loss": 0.892,
+      "step": 5795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0796562514673228e-05,
+      "loss": 0.9385,
+      "step": 5796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079382625935582e-05,
+      "loss": 0.6672,
+      "step": 5797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791089944225684e-05,
+      "loss": 0.9085,
+      "step": 5798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0788353569488992e-05,
+      "loss": 0.9031,
+      "step": 5799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785617135351927e-05,
+      "loss": 0.8974,
+      "step": 5800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078288064202067e-05,
+      "loss": 0.8981,
+      "step": 5801
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078014408970141e-05,
+      "loss": 0.8617,
+      "step": 5802
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0777407478600334e-05,
+      "loss": 0.8731,
+      "step": 5803
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0774670808923649e-05,
+      "loss": 0.9363,
+      "step": 5804
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0771934080877542e-05,
+      "loss": 0.9014,
+      "step": 5805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0769197294668228e-05,
+      "loss": 0.9263,
+      "step": 5806
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0766460450501915e-05,
+      "loss": 0.8987,
+      "step": 5807
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0763723548584813e-05,
+      "loss": 0.858,
+      "step": 5808
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760986589123145e-05,
+      "loss": 0.9591,
+      "step": 5809
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0758249572323134e-05,
+      "loss": 0.8509,
+      "step": 5810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0755512498391003e-05,
+      "loss": 0.9158,
+      "step": 5811
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0752775367532988e-05,
+      "loss": 0.9327,
+      "step": 5812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075003817995532e-05,
+      "loss": 0.519,
+      "step": 5813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 0.9359,
+      "step": 5814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744563635466e-05,
+      "loss": 0.8745,
+      "step": 5815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0741826278966836e-05,
+      "loss": 0.8901,
+      "step": 5816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0739088866573009e-05,
+      "loss": 0.8903,
+      "step": 5817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0736351398490772e-05,
+      "loss": 0.9084,
+      "step": 5818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0733613874926385e-05,
+      "loss": 0.9393,
+      "step": 5819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0730876296086118e-05,
+      "loss": 0.9081,
+      "step": 5820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728138662176237e-05,
+      "loss": 0.5745,
+      "step": 5821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0725400973403018e-05,
+      "loss": 0.4748,
+      "step": 5822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722663229972734e-05,
+      "loss": 0.9614,
+      "step": 5823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0719925432091671e-05,
+      "loss": 0.9187,
+      "step": 5824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0717187579966118e-05,
+      "loss": 0.889,
+      "step": 5825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714449673802356e-05,
+      "loss": 0.941,
+      "step": 5826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0711711713806684e-05,
+      "loss": 0.8704,
+      "step": 5827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07089737001854e-05,
+      "loss": 0.9009,
+      "step": 5828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0706235633144805e-05,
+      "loss": 0.8759,
+      "step": 5829
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070349751289121e-05,
+      "loss": 0.9175,
+      "step": 5830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070075933963092e-05,
+      "loss": 0.9362,
+      "step": 5831
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0698021113570248e-05,
+      "loss": 0.8776,
+      "step": 5832
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0695282834915517e-05,
+      "loss": 0.9113,
+      "step": 5833
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0692544503873046e-05,
+      "loss": 0.8859,
+      "step": 5834
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689806120649164e-05,
+      "loss": 0.8293,
+      "step": 5835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0687067685450199e-05,
+      "loss": 0.8905,
+      "step": 5836
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0684329198482484e-05,
+      "loss": 0.8961,
+      "step": 5837
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0681590659952355e-05,
+      "loss": 0.4971,
+      "step": 5838
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.067885207006616e-05,
+      "loss": 0.9251,
+      "step": 5839
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0676113429030238e-05,
+      "loss": 0.8968,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0673374737050941e-05,
+      "loss": 0.8835,
+      "step": 5841
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0670635994334626e-05,
+      "loss": 0.9581,
+      "step": 5842
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0667897201087644e-05,
+      "loss": 0.9049,
+      "step": 5843
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0665158357516358e-05,
+      "loss": 0.9111,
+      "step": 5844
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0662419463827136e-05,
+      "loss": 0.8586,
+      "step": 5845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0659680520226346e-05,
+      "loss": 0.8941,
+      "step": 5846
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0656941526920352e-05,
+      "loss": 0.9127,
+      "step": 5847
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.065420248411554e-05,
+      "loss": 0.8798,
+      "step": 5848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.065146339201828e-05,
+      "loss": 0.8721,
+      "step": 5849
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0648724250834966e-05,
+      "loss": 0.8811,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0645985060771978e-05,
+      "loss": 0.9536,
+      "step": 5851
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064324582203571e-05,
+      "loss": 0.9337,
+      "step": 5852
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0640506534832555e-05,
+      "loss": 0.8832,
+      "step": 5853
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0637767199368911e-05,
+      "loss": 0.9257,
+      "step": 5854
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0635027815851182e-05,
+      "loss": 0.953,
+      "step": 5855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632288384485773e-05,
+      "loss": 0.9053,
+      "step": 5856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062954890547909e-05,
+      "loss": 0.8971,
+      "step": 5857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0626809379037546e-05,
+      "loss": 0.8452,
+      "step": 5858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0624069805367558e-05,
+      "loss": 0.9254,
+      "step": 5859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0621330184675547e-05,
+      "loss": 0.9207,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0618590517167934e-05,
+      "loss": 0.8841,
+      "step": 5861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0615850803051149e-05,
+      "loss": 0.9402,
+      "step": 5862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613111042531618e-05,
+      "loss": 0.9183,
+      "step": 5863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610371235815777e-05,
+      "loss": 0.9006,
+      "step": 5864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0607631383110066e-05,
+      "loss": 0.9079,
+      "step": 5865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060489148462092e-05,
+      "loss": 0.9087,
+      "step": 5866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060215154055479e-05,
+      "loss": 0.844,
+      "step": 5867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0599411551118117e-05,
+      "loss": 0.5801,
+      "step": 5868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0596671516517356e-05,
+      "loss": 0.923,
+      "step": 5869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059393143695896e-05,
+      "loss": 0.8942,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0591191312649383e-05,
+      "loss": 0.6336,
+      "step": 5871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0588451143795093e-05,
+      "loss": 0.9095,
+      "step": 5872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0585710930602553e-05,
+      "loss": 0.9443,
+      "step": 5873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0582970673278227e-05,
+      "loss": 0.8963,
+      "step": 5874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0580230372028593e-05,
+      "loss": 0.883,
+      "step": 5875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0577490027060118e-05,
+      "loss": 0.9023,
+      "step": 5876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0574749638579287e-05,
+      "loss": 0.9146,
+      "step": 5877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0572009206792575e-05,
+      "loss": 0.9296,
+      "step": 5878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0569268731906468e-05,
+      "loss": 0.8934,
+      "step": 5879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0566528214127454e-05,
+      "loss": 0.8863,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0563787653662025e-05,
+      "loss": 0.9655,
+      "step": 5881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0561047050716673e-05,
+      "loss": 0.9215,
+      "step": 5882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0558306405497897e-05,
+      "loss": 0.8544,
+      "step": 5883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0555565718212198e-05,
+      "loss": 0.9151,
+      "step": 5884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552824989066075e-05,
+      "loss": 0.9251,
+      "step": 5885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055008421826604e-05,
+      "loss": 0.8658,
+      "step": 5886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0547343406018602e-05,
+      "loss": 0.9431,
+      "step": 5887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0544602552530274e-05,
+      "loss": 0.9187,
+      "step": 5888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054186165800757e-05,
+      "loss": 0.8959,
+      "step": 5889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053912072265701e-05,
+      "loss": 0.9885,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0536379746685117e-05,
+      "loss": 0.9169,
+      "step": 5891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0533638730298415e-05,
+      "loss": 0.8738,
+      "step": 5892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530897673703431e-05,
+      "loss": 0.9496,
+      "step": 5893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0528156577106703e-05,
+      "loss": 0.9177,
+      "step": 5894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052541544071476e-05,
+      "loss": 0.972,
+      "step": 5895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0522674264734141e-05,
+      "loss": 0.9083,
+      "step": 5896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0519933049371385e-05,
+      "loss": 0.9448,
+      "step": 5897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051719179483304e-05,
+      "loss": 0.938,
+      "step": 5898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0514450501325646e-05,
+      "loss": 0.8511,
+      "step": 5899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0511709169055758e-05,
+      "loss": 0.8917,
+      "step": 5900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0508967798229921e-05,
+      "loss": 0.87,
+      "step": 5901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0506226389054697e-05,
+      "loss": 0.8872,
+      "step": 5902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0503484941736641e-05,
+      "loss": 0.9236,
+      "step": 5903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0500743456482313e-05,
+      "loss": 0.8835,
+      "step": 5904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049800193349828e-05,
+      "loss": 0.9194,
+      "step": 5905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0495260372991107e-05,
+      "loss": 0.559,
+      "step": 5906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0492518775167364e-05,
+      "loss": 0.8907,
+      "step": 5907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0489777140233619e-05,
+      "loss": 0.9127,
+      "step": 5908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0487035468396456e-05,
+      "loss": 0.8779,
+      "step": 5909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484293759862443e-05,
+      "loss": 0.9101,
+      "step": 5910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0481552014838164e-05,
+      "loss": 0.8341,
+      "step": 5911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0478810233530203e-05,
+      "loss": 0.9067,
+      "step": 5912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476068416145146e-05,
+      "loss": 0.9039,
+      "step": 5913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0473326562889583e-05,
+      "loss": 0.9238,
+      "step": 5914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0470584673970102e-05,
+      "loss": 0.9046,
+      "step": 5915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04678427495933e-05,
+      "loss": 0.9445,
+      "step": 5916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0465100789965774e-05,
+      "loss": 0.8329,
+      "step": 5917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0462358795294123e-05,
+      "loss": 0.8776,
+      "step": 5918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045961676578495e-05,
+      "loss": 0.8925,
+      "step": 5919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0456874701644857e-05,
+      "loss": 0.9204,
+      "step": 5920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0454132603080451e-05,
+      "loss": 0.9366,
+      "step": 5921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0451390470298346e-05,
+      "loss": 0.9198,
+      "step": 5922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044864830350515e-05,
+      "loss": 0.8792,
+      "step": 5923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0445906102907482e-05,
+      "loss": 0.8769,
+      "step": 5924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0443163868711958e-05,
+      "loss": 0.9322,
+      "step": 5925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440421601125196e-05,
+      "loss": 0.8701,
+      "step": 5926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0437679300353825e-05,
+      "loss": 0.8497,
+      "step": 5927
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0434936966604462e-05,
+      "loss": 0.9364,
+      "step": 5928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043219460008374e-05,
+      "loss": 0.8916,
+      "step": 5929
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0429452200998292e-05,
+      "loss": 0.9094,
+      "step": 5930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0426709769554741e-05,
+      "loss": 0.9571,
+      "step": 5931
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423967305959727e-05,
+      "loss": 0.9048,
+      "step": 5932
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042122481041989e-05,
+      "loss": 0.9371,
+      "step": 5933
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0418482283141867e-05,
+      "loss": 0.8713,
+      "step": 5934
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0415739724332301e-05,
+      "loss": 0.8904,
+      "step": 5935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0412997134197837e-05,
+      "loss": 0.8945,
+      "step": 5936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0410254512945118e-05,
+      "loss": 0.9329,
+      "step": 5937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407511860780798e-05,
+      "loss": 0.8848,
+      "step": 5938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0404769177911531e-05,
+      "loss": 0.9487,
+      "step": 5939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402026464543969e-05,
+      "loss": 0.9196,
+      "step": 5940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0399283720884761e-05,
+      "loss": 0.9376,
+      "step": 5941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0396540947140573e-05,
+      "loss": 0.9305,
+      "step": 5942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0393798143518066e-05,
+      "loss": 0.862,
+      "step": 5943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0391055310223899e-05,
+      "loss": 0.9767,
+      "step": 5944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0388312447464739e-05,
+      "loss": 0.9292,
+      "step": 5945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0385569555447257e-05,
+      "loss": 0.8943,
+      "step": 5946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0382826634378115e-05,
+      "loss": 0.9033,
+      "step": 5947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0380083684463993e-05,
+      "loss": 0.8826,
+      "step": 5948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037734070591156e-05,
+      "loss": 0.666,
+      "step": 5949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0374597698927496e-05,
+      "loss": 0.8568,
+      "step": 5950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371854663718477e-05,
+      "loss": 0.8783,
+      "step": 5951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0369111600491184e-05,
+      "loss": 0.8709,
+      "step": 5952
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366368509452298e-05,
+      "loss": 0.5811,
+      "step": 5953
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0363625390808505e-05,
+      "loss": 0.8944,
+      "step": 5954
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0360882244766493e-05,
+      "loss": 0.9004,
+      "step": 5955
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0358139071532949e-05,
+      "loss": 0.5351,
+      "step": 5956
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355395871314563e-05,
+      "loss": 0.914,
+      "step": 5957
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0352652644318033e-05,
+      "loss": 0.91,
+      "step": 5958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0349909390750046e-05,
+      "loss": 0.8506,
+      "step": 5959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0347166110817305e-05,
+      "loss": 0.9256,
+      "step": 5960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0344422804726513e-05,
+      "loss": 0.89,
+      "step": 5961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.034167947268436e-05,
+      "loss": 0.9264,
+      "step": 5962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0338936114897558e-05,
+      "loss": 0.8549,
+      "step": 5963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0336192731572805e-05,
+      "loss": 0.9311,
+      "step": 5964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0333449322916812e-05,
+      "loss": 0.9059,
+      "step": 5965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0330705889136288e-05,
+      "loss": 0.9334,
+      "step": 5966
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0327962430437938e-05,
+      "loss": 0.8604,
+      "step": 5967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0325218947028483e-05,
+      "loss": 0.9616,
+      "step": 5968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.032247543911463e-05,
+      "loss": 0.897,
+      "step": 5969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0319731906903098e-05,
+      "loss": 0.8634,
+      "step": 5970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0316988350600608e-05,
+      "loss": 0.9492,
+      "step": 5971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0314244770413875e-05,
+      "loss": 0.9027,
+      "step": 5972
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0311501166549622e-05,
+      "loss": 0.8784,
+      "step": 5973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0308757539214573e-05,
+      "loss": 0.868,
+      "step": 5974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0306013888615451e-05,
+      "loss": 0.9261,
+      "step": 5975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0303270214958988e-05,
+      "loss": 0.9409,
+      "step": 5976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0300526518451906e-05,
+      "loss": 0.8906,
+      "step": 5977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0297782799300939e-05,
+      "loss": 0.9074,
+      "step": 5978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0295039057712819e-05,
+      "loss": 0.8954,
+      "step": 5979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0292295293894279e-05,
+      "loss": 0.8543,
+      "step": 5980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0289551508052054e-05,
+      "loss": 0.8912,
+      "step": 5981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0286807700392884e-05,
+      "loss": 0.8934,
+      "step": 5982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0284063871123504e-05,
+      "loss": 0.9107,
+      "step": 5983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0281320020450657e-05,
+      "loss": 0.9253,
+      "step": 5984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278576148581083e-05,
+      "loss": 0.4727,
+      "step": 5985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0275832255721527e-05,
+      "loss": 0.92,
+      "step": 5986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0273088342078734e-05,
+      "loss": 0.8936,
+      "step": 5987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027034440785945e-05,
+      "loss": 0.9086,
+      "step": 5988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0267600453270422e-05,
+      "loss": 0.9016,
+      "step": 5989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02648564785184e-05,
+      "loss": 0.8912,
+      "step": 5990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0262112483810141e-05,
+      "loss": 0.8625,
+      "step": 5991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025936846935239e-05,
+      "loss": 0.9536,
+      "step": 5992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0256624435351909e-05,
+      "loss": 0.9049,
+      "step": 5993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253880382015446e-05,
+      "loss": 0.9591,
+      "step": 5994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0251136309549764e-05,
+      "loss": 0.9245,
+      "step": 5995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0248392218161616e-05,
+      "loss": 0.8644,
+      "step": 5996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0245648108057767e-05,
+      "loss": 0.9022,
+      "step": 5997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0242903979444976e-05,
+      "loss": 0.8957,
+      "step": 5998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0240159832530007e-05,
+      "loss": 0.8679,
+      "step": 5999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0237415667519626e-05,
+      "loss": 0.8834,
+      "step": 6000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0234671484620595e-05,
+      "loss": 0.9459,
+      "step": 6001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0231927284039685e-05,
+      "loss": 0.9198,
+      "step": 6002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229183065983659e-05,
+      "loss": 0.9313,
+      "step": 6003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022643883065929e-05,
+      "loss": 0.5398,
+      "step": 6004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022369457827335e-05,
+      "loss": 0.8798,
+      "step": 6005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0220950309032611e-05,
+      "loss": 0.899,
+      "step": 6006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0218206023143843e-05,
+      "loss": 0.5092,
+      "step": 6007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0215461720813824e-05,
+      "loss": 0.8795,
+      "step": 6008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021271740224933e-05,
+      "loss": 0.9095,
+      "step": 6009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0209973067657138e-05,
+      "loss": 0.8814,
+      "step": 6010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0207228717244023e-05,
+      "loss": 0.9164,
+      "step": 6011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0204484351216768e-05,
+      "loss": 0.9053,
+      "step": 6012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0201739969782154e-05,
+      "loss": 0.9138,
+      "step": 6013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0198995573146964e-05,
+      "loss": 0.9016,
+      "step": 6014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0196251161517975e-05,
+      "loss": 0.8904,
+      "step": 6015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019350673510198e-05,
+      "loss": 0.9281,
+      "step": 6016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0190762294105757e-05,
+      "loss": 0.8883,
+      "step": 6017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188017838736095e-05,
+      "loss": 0.8952,
+      "step": 6018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0185273369199781e-05,
+      "loss": 0.9215,
+      "step": 6019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0182528885703606e-05,
+      "loss": 0.8645,
+      "step": 6020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0179784388454356e-05,
+      "loss": 0.8885,
+      "step": 6021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177039877658825e-05,
+      "loss": 0.8849,
+      "step": 6022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174295353523805e-05,
+      "loss": 0.9749,
+      "step": 6023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0171550816256084e-05,
+      "loss": 0.8529,
+      "step": 6024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0168806266062459e-05,
+      "loss": 0.9517,
+      "step": 6025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166061703149726e-05,
+      "loss": 0.8818,
+      "step": 6026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0163317127724677e-05,
+      "loss": 0.9011,
+      "step": 6027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0160572539994111e-05,
+      "loss": 0.9153,
+      "step": 6028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0157827940164826e-05,
+      "loss": 0.9346,
+      "step": 6029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155083328443617e-05,
+      "loss": 0.8783,
+      "step": 6030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0152338705037288e-05,
+      "loss": 0.8539,
+      "step": 6031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149594070152638e-05,
+      "loss": 0.878,
+      "step": 6032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0146849423996463e-05,
+      "loss": 0.8729,
+      "step": 6033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144104766775574e-05,
+      "loss": 0.8846,
+      "step": 6034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0141360098696767e-05,
+      "loss": 0.9015,
+      "step": 6035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0138615419966844e-05,
+      "loss": 0.8552,
+      "step": 6036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135870730792614e-05,
+      "loss": 0.947,
+      "step": 6037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0133126031380882e-05,
+      "loss": 0.9075,
+      "step": 6038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130381321938451e-05,
+      "loss": 0.927,
+      "step": 6039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0127636602672129e-05,
+      "loss": 0.9367,
+      "step": 6040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0124891873788725e-05,
+      "loss": 0.9216,
+      "step": 6041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0122147135495044e-05,
+      "loss": 0.9313,
+      "step": 6042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0119402387997896e-05,
+      "loss": 0.8972,
+      "step": 6043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116657631504094e-05,
+      "loss": 0.8653,
+      "step": 6044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0113912866220443e-05,
+      "loss": 0.9084,
+      "step": 6045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0111168092353755e-05,
+      "loss": 0.9032,
+      "step": 6046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0108423310110847e-05,
+      "loss": 0.8754,
+      "step": 6047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105678519698524e-05,
+      "loss": 0.8729,
+      "step": 6048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01029337213236e-05,
+      "loss": 0.8324,
+      "step": 6049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0100188915192891e-05,
+      "loss": 0.8896,
+      "step": 6050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009744410151321e-05,
+      "loss": 0.9343,
+      "step": 6051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0094699280491371e-05,
+      "loss": 0.9456,
+      "step": 6052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0091954452334191e-05,
+      "loss": 0.9061,
+      "step": 6053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0089209617248483e-05,
+      "loss": 0.8798,
+      "step": 6054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0086464775441064e-05,
+      "loss": 0.9315,
+      "step": 6055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0083719927118754e-05,
+      "loss": 0.8658,
+      "step": 6056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0080975072488364e-05,
+      "loss": 0.94,
+      "step": 6057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0078230211756714e-05,
+      "loss": 0.8927,
+      "step": 6058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0075485345130624e-05,
+      "loss": 0.8835,
+      "step": 6059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007274047281691e-05,
+      "loss": 0.8669,
+      "step": 6060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069995595022393e-05,
+      "loss": 0.8679,
+      "step": 6061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006725071195389e-05,
+      "loss": 0.899,
+      "step": 6062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0064505823818222e-05,
+      "loss": 0.8759,
+      "step": 6063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006176093082221e-05,
+      "loss": 0.8917,
+      "step": 6064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059016033172675e-05,
+      "loss": 0.8805,
+      "step": 6065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056271131076436e-05,
+      "loss": 0.9238,
+      "step": 6066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0053526224740313e-05,
+      "loss": 0.8693,
+      "step": 6067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050781314371132e-05,
+      "loss": 0.8795,
+      "step": 6068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004803640017571e-05,
+      "loss": 0.8863,
+      "step": 6069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045291482360871e-05,
+      "loss": 0.9333,
+      "step": 6070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042546561133439e-05,
+      "loss": 0.8839,
+      "step": 6071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039801636700234e-05,
+      "loss": 0.9111,
+      "step": 6072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003705670926808e-05,
+      "loss": 0.8934,
+      "step": 6073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0034311779043801e-05,
+      "loss": 0.8698,
+      "step": 6074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003156684623422e-05,
+      "loss": 0.8802,
+      "step": 6075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028821911046158e-05,
+      "loss": 0.9492,
+      "step": 6076
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0026076973686443e-05,
+      "loss": 0.9132,
+      "step": 6077
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0023332034361894e-05,
+      "loss": 0.9114,
+      "step": 6078
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0020587093279339e-05,
+      "loss": 0.9334,
+      "step": 6079
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.00178421506456e-05,
+      "loss": 0.9254,
+      "step": 6080
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0015097206667501e-05,
+      "loss": 0.9315,
+      "step": 6081
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0012352261551868e-05,
+      "loss": 0.8624,
+      "step": 6082
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0009607315505523e-05,
+      "loss": 0.9658,
+      "step": 6083
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0006862368735292e-05,
+      "loss": 0.9177,
+      "step": 6084
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0004117421448e-05,
+      "loss": 0.8756,
+      "step": 6085
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0001372473850472e-05,
+      "loss": 0.8567,
+      "step": 6086
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.998627526149531e-06,
+      "loss": 0.8946,
+      "step": 6087
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995882578552002e-06,
+      "loss": 0.8775,
+      "step": 6088
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99313763126471e-06,
+      "loss": 0.8695,
+      "step": 6089
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99039268449448e-06,
+      "loss": 0.9,
+      "step": 6090
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987647738448134e-06,
+      "loss": 0.8594,
+      "step": 6091
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.984902793332502e-06,
+      "loss": 0.9054,
+      "step": 6092
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982157849354405e-06,
+      "loss": 0.9,
+      "step": 6093
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982157849354405e-06,
+      "loss": 0.8465,
+      "step": 6094
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979412906720663e-06,
+      "loss": 0.9254,
+      "step": 6095
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976667965638111e-06,
+      "loss": 0.9485,
+      "step": 6096
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97392302631356e-06,
+      "loss": 0.9689,
+      "step": 6097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971178088953845e-06,
+      "loss": 0.9346,
+      "step": 6098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968433153765782e-06,
+      "loss": 0.9114,
+      "step": 6099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.965688220956202e-06,
+      "loss": 0.9382,
+      "step": 6100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96294329073192e-06,
+      "loss": 0.8961,
+      "step": 6101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.96019836329977e-06,
+      "loss": 0.9227,
+      "step": 6102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.957453438866563e-06,
+      "loss": 0.9177,
+      "step": 6103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95470851763913e-06,
+      "loss": 0.9208,
+      "step": 6104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.951963599824294e-06,
+      "loss": 0.887,
+      "step": 6105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949218685628871e-06,
+      "loss": 0.8856,
+      "step": 6106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94647377525969e-06,
+      "loss": 0.8956,
+      "step": 6107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943728868923568e-06,
+      "loss": 0.8843,
+      "step": 6108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940983966827328e-06,
+      "loss": 0.872,
+      "step": 6109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.938239069177792e-06,
+      "loss": 0.837,
+      "step": 6110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.935494176181781e-06,
+      "loss": 0.8946,
+      "step": 6111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.932749288046113e-06,
+      "loss": 0.8783,
+      "step": 6112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93000440497761e-06,
+      "loss": 0.9159,
+      "step": 6113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.927259527183095e-06,
+      "loss": 0.9161,
+      "step": 6114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.924514654869379e-06,
+      "loss": 0.9147,
+      "step": 6115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.921769788243291e-06,
+      "loss": 0.9115,
+      "step": 6116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91902492751164e-06,
+      "loss": 0.9427,
+      "step": 6117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.916280072881251e-06,
+      "loss": 0.5059,
+      "step": 6118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.913535224558936e-06,
+      "loss": 0.8606,
+      "step": 6119
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91079038275152e-06,
+      "loss": 0.9344,
+      "step": 6120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.908045547665809e-06,
+      "loss": 0.882,
+      "step": 6121
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90530071950863e-06,
+      "loss": 0.861,
+      "step": 6122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90255589848679e-06,
+      "loss": 0.9402,
+      "step": 6123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.899811084807112e-06,
+      "loss": 0.911,
+      "step": 6124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.897066278676405e-06,
+      "loss": 0.9069,
+      "step": 6125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894321480301481e-06,
+      "loss": 0.8631,
+      "step": 6126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89157668988916e-06,
+      "loss": 0.9015,
+      "step": 6127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.888831907646246e-06,
+      "loss": 0.8231,
+      "step": 6128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88608713377956e-06,
+      "loss": 0.8905,
+      "step": 6129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88334236849591e-06,
+      "loss": 0.9178,
+      "step": 6130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880597612002106e-06,
+      "loss": 0.929,
+      "step": 6131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87785286450496e-06,
+      "loss": 0.8784,
+      "step": 6132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875108126211279e-06,
+      "loss": 0.8909,
+      "step": 6133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872363397327873e-06,
+      "loss": 0.8936,
+      "step": 6134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869618678061552e-06,
+      "loss": 0.9433,
+      "step": 6135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.866873968619123e-06,
+      "loss": 0.9187,
+      "step": 6136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864129269207388e-06,
+      "loss": 0.8963,
+      "step": 6137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86138458003316e-06,
+      "loss": 0.9101,
+      "step": 6138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.858639901303237e-06,
+      "loss": 0.8998,
+      "step": 6139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.855895233224431e-06,
+      "loss": 0.8911,
+      "step": 6140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.853150576003537e-06,
+      "loss": 0.9386,
+      "step": 6141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850405929847367e-06,
+      "loss": 0.8668,
+      "step": 6142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847661294962712e-06,
+      "loss": 0.8685,
+      "step": 6143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844916671556384e-06,
+      "loss": 0.902,
+      "step": 6144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842172059835175e-06,
+      "loss": 0.9193,
+      "step": 6145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839427460005892e-06,
+      "loss": 0.8603,
+      "step": 6146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.836682872275328e-06,
+      "loss": 0.9421,
+      "step": 6147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833938296850279e-06,
+      "loss": 0.8536,
+      "step": 6148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.831193733937546e-06,
+      "loss": 0.9063,
+      "step": 6149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82844918374392e-06,
+      "loss": 0.8937,
+      "step": 6150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8257046464762e-06,
+      "loss": 0.8659,
+      "step": 6151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822960122341178e-06,
+      "loss": 0.9107,
+      "step": 6152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.820215611545645e-06,
+      "loss": 0.9363,
+      "step": 6153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.817471114296398e-06,
+      "loss": 0.8773,
+      "step": 6154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81472663080022e-06,
+      "loss": 0.8961,
+      "step": 6155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81198216126391e-06,
+      "loss": 0.9188,
+      "step": 6156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.809237705894246e-06,
+      "loss": 0.8572,
+      "step": 6157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806493264898025e-06,
+      "loss": 0.8381,
+      "step": 6158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.803748838482026e-06,
+      "loss": 0.9032,
+      "step": 6159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80100442685304e-06,
+      "loss": 0.9019,
+      "step": 6160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.798260030217846e-06,
+      "loss": 0.9023,
+      "step": 6161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.795515648783235e-06,
+      "loss": 0.8676,
+      "step": 6162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792771282755979e-06,
+      "loss": 0.9076,
+      "step": 6163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.790026932342867e-06,
+      "loss": 0.9017,
+      "step": 6164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78728259775067e-06,
+      "loss": 0.9143,
+      "step": 6165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.784538279186178e-06,
+      "loss": 0.8848,
+      "step": 6166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.781793976856162e-06,
+      "loss": 0.8898,
+      "step": 6167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.779049690967392e-06,
+      "loss": 0.9329,
+      "step": 6168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.776305421726654e-06,
+      "loss": 0.915,
+      "step": 6169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.773561169340711e-06,
+      "loss": 0.872,
+      "step": 6170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770816934016346e-06,
+      "loss": 0.9039,
+      "step": 6171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.768072715960318e-06,
+      "loss": 0.9136,
+      "step": 6172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.765328515379407e-06,
+      "loss": 0.9085,
+      "step": 6173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.762584332480378e-06,
+      "loss": 0.8635,
+      "step": 6174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.759840167469995e-06,
+      "loss": 0.9255,
+      "step": 6175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.757096020555026e-06,
+      "loss": 0.9158,
+      "step": 6176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.754351891942236e-06,
+      "loss": 0.9279,
+      "step": 6177
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751607781838389e-06,
+      "loss": 0.969,
+      "step": 6178
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.74886369045024e-06,
+      "loss": 0.8959,
+      "step": 6179
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746119617984559e-06,
+      "loss": 0.9182,
+      "step": 6180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.743375564648092e-06,
+      "loss": 0.9229,
+      "step": 6181
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740631530647611e-06,
+      "loss": 0.9114,
+      "step": 6182
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73788751618986e-06,
+      "loss": 0.8696,
+      "step": 6183
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735143521481601e-06,
+      "loss": 0.5304,
+      "step": 6184
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.732399546729578e-06,
+      "loss": 0.9456,
+      "step": 6185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.729655592140555e-06,
+      "loss": 0.8579,
+      "step": 6186
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.726911657921268e-06,
+      "loss": 0.8835,
+      "step": 6187
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.724167744278475e-06,
+      "loss": 0.9313,
+      "step": 6188
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.721423851418922e-06,
+      "loss": 0.9427,
+      "step": 6189
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.718679979549346e-06,
+      "loss": 0.9163,
+      "step": 6190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715936128876501e-06,
+      "loss": 0.8808,
+      "step": 6191
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713192299607119e-06,
+      "loss": 0.8622,
+      "step": 6192
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710448491947951e-06,
+      "loss": 0.8756,
+      "step": 6193
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.707704706105724e-06,
+      "loss": 0.9184,
+      "step": 6194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.704960942287184e-06,
+      "loss": 0.8621,
+      "step": 6195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.702217200699065e-06,
+      "loss": 0.8629,
+      "step": 6196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.699473481548097e-06,
+      "loss": 0.8756,
+      "step": 6197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.696729785041019e-06,
+      "loss": 0.9433,
+      "step": 6198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.693986111384552e-06,
+      "loss": 0.9006,
+      "step": 6199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.691242460785433e-06,
+      "loss": 0.9019,
+      "step": 6200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.688498833450381e-06,
+      "loss": 0.8941,
+      "step": 6201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.685755229586129e-06,
+      "loss": 0.926,
+      "step": 6202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.683011649399393e-06,
+      "loss": 0.9137,
+      "step": 6203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.680268093096904e-06,
+      "loss": 0.8896,
+      "step": 6204
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.67752456088537e-06,
+      "loss": 0.8628,
+      "step": 6205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.67478105297152e-06,
+      "loss": 0.9319,
+      "step": 6206
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.672037569562062e-06,
+      "loss": 0.893,
+      "step": 6207
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669294110863717e-06,
+      "loss": 0.9099,
+      "step": 6208
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.666550677083193e-06,
+      "loss": 0.8877,
+      "step": 6209
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663807268427197e-06,
+      "loss": 0.9055,
+      "step": 6210
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.661063885102447e-06,
+      "loss": 0.9017,
+      "step": 6211
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658320527315642e-06,
+      "loss": 0.9172,
+      "step": 6212
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.655577195273492e-06,
+      "loss": 0.904,
+      "step": 6213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.652833889182695e-06,
+      "loss": 0.8784,
+      "step": 6214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650090609249957e-06,
+      "loss": 0.8969,
+      "step": 6215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64734735568197e-06,
+      "loss": 0.5931,
+      "step": 6216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644604128685438e-06,
+      "loss": 0.8923,
+      "step": 6217
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.641860928467054e-06,
+      "loss": 0.9398,
+      "step": 6218
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63911775523351e-06,
+      "loss": 0.9101,
+      "step": 6219
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6363746091915e-06,
+      "loss": 0.8815,
+      "step": 6220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633631490547705e-06,
+      "loss": 0.8735,
+      "step": 6221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63088839950882e-06,
+      "loss": 0.8746,
+      "step": 6222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628145336281524e-06,
+      "loss": 0.912,
+      "step": 6223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625402301072508e-06,
+      "loss": 0.884,
+      "step": 6224
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62265929408844e-06,
+      "loss": 0.8645,
+      "step": 6225
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61991631553601e-06,
+      "loss": 0.5233,
+      "step": 6226
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.617173365621885e-06,
+      "loss": 0.8633,
+      "step": 6227
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.614430444552746e-06,
+      "loss": 0.9161,
+      "step": 6228
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61168755253526e-06,
+      "loss": 0.917,
+      "step": 6229
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.608944689776104e-06,
+      "loss": 0.8664,
+      "step": 6230
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.60620185648194e-06,
+      "loss": 0.8552,
+      "step": 6231
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.603459052859428e-06,
+      "loss": 0.8766,
+      "step": 6232
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.600716279115244e-06,
+      "loss": 0.6077,
+      "step": 6233
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597973535456036e-06,
+      "loss": 0.8715,
+      "step": 6234
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.595230822088472e-06,
+      "loss": 0.9059,
+      "step": 6235
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5924881392192e-06,
+      "loss": 0.9261,
+      "step": 6236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.589745487054884e-06,
+      "loss": 0.856,
+      "step": 6237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.587002865802166e-06,
+      "loss": 0.8793,
+      "step": 6238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.584260275667702e-06,
+      "loss": 0.8615,
+      "step": 6239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.581517716858137e-06,
+      "loss": 0.9255,
+      "step": 6240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.578775189580113e-06,
+      "loss": 0.9236,
+      "step": 6241
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576032694040278e-06,
+      "loss": 0.9182,
+      "step": 6242
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.573290230445262e-06,
+      "loss": 0.9057,
+      "step": 6243
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570547799001715e-06,
+      "loss": 0.8626,
+      "step": 6244
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56780539991626e-06,
+      "loss": 0.898,
+      "step": 6245
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56506303339554e-06,
+      "loss": 0.8837,
+      "step": 6246
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.562320699646177e-06,
+      "loss": 0.465,
+      "step": 6247
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559578398874805e-06,
+      "loss": 0.8691,
+      "step": 6248
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.556836131288043e-06,
+      "loss": 0.8764,
+      "step": 6249
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55409389709252e-06,
+      "loss": 0.9064,
+      "step": 6250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551351696494854e-06,
+      "loss": 0.8783,
+      "step": 6251
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.548609529701657e-06,
+      "loss": 0.8816,
+      "step": 6252
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545867396919554e-06,
+      "loss": 0.9174,
+      "step": 6253
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.543125298355147e-06,
+      "loss": 0.536,
+      "step": 6254
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.540383234215057e-06,
+      "loss": 0.9143,
+      "step": 6255
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537641204705878e-06,
+      "loss": 0.8863,
+      "step": 6256
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53489921003423e-06,
+      "loss": 0.9302,
+      "step": 6257
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.532157250406701e-06,
+      "loss": 0.8985,
+      "step": 6258
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529415326029902e-06,
+      "loss": 0.9062,
+      "step": 6259
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52667343711042e-06,
+      "loss": 0.8402,
+      "step": 6260
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523931583854856e-06,
+      "loss": 0.9418,
+      "step": 6261
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.521189766469802e-06,
+      "loss": 0.8885,
+      "step": 6262
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51844798516184e-06,
+      "loss": 0.9017,
+      "step": 6263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515706240137563e-06,
+      "loss": 0.8898,
+      "step": 6264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.512964531603548e-06,
+      "loss": 0.8756,
+      "step": 6265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.510222859766383e-06,
+      "loss": 0.8845,
+      "step": 6266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507481224832637e-06,
+      "loss": 0.9251,
+      "step": 6267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.504739627008894e-06,
+      "loss": 0.9047,
+      "step": 6268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501998066501718e-06,
+      "loss": 0.8997,
+      "step": 6269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499256543517688e-06,
+      "loss": 0.9141,
+      "step": 6270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.496515058263358e-06,
+      "loss": 0.9242,
+      "step": 6271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.493773610945305e-06,
+      "loss": 0.9115,
+      "step": 6272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.491032201770082e-06,
+      "loss": 0.9039,
+      "step": 6273
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488290830944247e-06,
+      "loss": 0.8778,
+      "step": 6274
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485549498674357e-06,
+      "loss": 0.8645,
+      "step": 6275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482808205166962e-06,
+      "loss": 0.886,
+      "step": 6276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.480066950628619e-06,
+      "loss": 0.9098,
+      "step": 6277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47732573526586e-06,
+      "loss": 0.8562,
+      "step": 6278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.474584559285244e-06,
+      "loss": 0.9459,
+      "step": 6279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.471843422893299e-06,
+      "loss": 0.8966,
+      "step": 6280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46910232629657e-06,
+      "loss": 0.8854,
+      "step": 6281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46636126970159e-06,
+      "loss": 0.8805,
+      "step": 6282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.463620253314886e-06,
+      "loss": 0.9071,
+      "step": 6283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.460879277342995e-06,
+      "loss": 0.8617,
+      "step": 6284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458138341992433e-06,
+      "loss": 0.8949,
+      "step": 6285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.455397447469731e-06,
+      "loss": 0.8953,
+      "step": 6286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.452656593981398e-06,
+      "loss": 0.8699,
+      "step": 6287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.449915781733961e-06,
+      "loss": 0.9029,
+      "step": 6288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447175010933925e-06,
+      "loss": 0.9376,
+      "step": 6289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.444434281787806e-06,
+      "loss": 0.9016,
+      "step": 6290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.441693594502104e-06,
+      "loss": 0.8601,
+      "step": 6291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43895294928333e-06,
+      "loss": 0.8494,
+      "step": 6292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.436212346337981e-06,
+      "loss": 0.918,
+      "step": 6293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.433471785872548e-06,
+      "loss": 0.8758,
+      "step": 6294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430731268093537e-06,
+      "loss": 0.9685,
+      "step": 6295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.427990793207428e-06,
+      "loss": 0.5764,
+      "step": 6296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.425250361420718e-06,
+      "loss": 0.9172,
+      "step": 6297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.422509972939881e-06,
+      "loss": 0.5064,
+      "step": 6298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41976962797141e-06,
+      "loss": 0.8845,
+      "step": 6299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41702932672177e-06,
+      "loss": 0.8893,
+      "step": 6300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41428906939745e-06,
+      "loss": 0.9128,
+      "step": 6301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.411548856204907e-06,
+      "loss": 0.9045,
+      "step": 6302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40880868735062e-06,
+      "loss": 0.8853,
+      "step": 6303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.406068563041047e-06,
+      "loss": 0.8627,
+      "step": 6304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.403328483482647e-06,
+      "loss": 0.8954,
+      "step": 6305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.400588448881888e-06,
+      "loss": 0.8958,
+      "step": 6306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397848459445212e-06,
+      "loss": 0.901,
+      "step": 6307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395108515379082e-06,
+      "loss": 0.9036,
+      "step": 6308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.392368616889936e-06,
+      "loss": 0.5158,
+      "step": 6309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389628764184225e-06,
+      "loss": 0.9003,
+      "step": 6310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386888957468383e-06,
+      "loss": 0.8762,
+      "step": 6311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.384149196948855e-06,
+      "loss": 0.9471,
+      "step": 6312
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.381409482832068e-06,
+      "loss": 0.9506,
+      "step": 6313
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.378669815324456e-06,
+      "loss": 0.9135,
+      "step": 6314
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.375930194632447e-06,
+      "loss": 0.8935,
+      "step": 6315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.373190620962459e-06,
+      "loss": 0.8636,
+      "step": 6316
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.370451094520915e-06,
+      "loss": 0.9301,
+      "step": 6317
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.36771161551423e-06,
+      "loss": 0.8771,
+      "step": 6318
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.364972184148821e-06,
+      "loss": 0.9044,
+      "step": 6319
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.362232800631087e-06,
+      "loss": 0.8845,
+      "step": 6320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.359493465167447e-06,
+      "loss": 0.8955,
+      "step": 6321
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.35675417796429e-06,
+      "loss": 0.8887,
+      "step": 6322
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.354014939228024e-06,
+      "loss": 0.8586,
+      "step": 6323
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.351275749165037e-06,
+      "loss": 0.8735,
+      "step": 6324
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.348536607981722e-06,
+      "loss": 0.864,
+      "step": 6325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345797515884466e-06,
+      "loss": 0.9393,
+      "step": 6326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34305847307965e-06,
+      "loss": 0.9261,
+      "step": 6327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34031947977366e-06,
+      "loss": 0.8799,
+      "step": 6328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.337580536172864e-06,
+      "loss": 0.8842,
+      "step": 6329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.334841642483643e-06,
+      "loss": 0.9374,
+      "step": 6330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332102798912357e-06,
+      "loss": 0.8893,
+      "step": 6331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.329364005665377e-06,
+      "loss": 0.8829,
+      "step": 6332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326625262949059e-06,
+      "loss": 0.9572,
+      "step": 6333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.323886570969764e-06,
+      "loss": 0.9013,
+      "step": 6334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321147929933847e-06,
+      "loss": 0.8461,
+      "step": 6335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.318409340047646e-06,
+      "loss": 0.8942,
+      "step": 6336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.315670801517523e-06,
+      "loss": 0.9008,
+      "step": 6337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.312932314549804e-06,
+      "loss": 0.8626,
+      "step": 6338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310193879350839e-06,
+      "loss": 0.8652,
+      "step": 6339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.307455496126954e-06,
+      "loss": 0.9184,
+      "step": 6340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.304717165084486e-06,
+      "loss": 0.5934,
+      "step": 6341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.301978886429752e-06,
+      "loss": 0.9075,
+      "step": 6342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299240660369085e-06,
+      "loss": 0.8898,
+      "step": 6343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.296502487108792e-06,
+      "loss": 0.9158,
+      "step": 6344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.293764366855196e-06,
+      "loss": 0.9169,
+      "step": 6345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.291026299814602e-06,
+      "loss": 0.8206,
+      "step": 6346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28828828619332e-06,
+      "loss": 0.9027,
+      "step": 6347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28555032619765e-06,
+      "loss": 0.9386,
+      "step": 6348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.282812420033886e-06,
+      "loss": 0.9155,
+      "step": 6349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28007456790833e-06,
+      "loss": 0.9051,
+      "step": 6350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277336770027266e-06,
+      "loss": 0.9156,
+      "step": 6351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.274599026596987e-06,
+      "loss": 0.8722,
+      "step": 6352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271861337823763e-06,
+      "loss": 0.896,
+      "step": 6353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.269123703913885e-06,
+      "loss": 0.916,
+      "step": 6354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266386125073613e-06,
+      "loss": 0.9016,
+      "step": 6355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263648601509231e-06,
+      "loss": 0.9495,
+      "step": 6356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.260911133426996e-06,
+      "loss": 0.8777,
+      "step": 6357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258173721033166e-06,
+      "loss": 0.8484,
+      "step": 6358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255436364534005e-06,
+      "loss": 0.8909,
+      "step": 6359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252699064135759e-06,
+      "loss": 0.9103,
+      "step": 6360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249961820044683e-06,
+      "loss": 0.8851,
+      "step": 6361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.247224632467014e-06,
+      "loss": 0.9554,
+      "step": 6362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.244487501608999e-06,
+      "loss": 0.8316,
+      "step": 6363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.241750427676866e-06,
+      "loss": 0.8223,
+      "step": 6364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239013410876856e-06,
+      "loss": 0.8726,
+      "step": 6365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236276451415189e-06,
+      "loss": 0.9158,
+      "step": 6366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23353954949809e-06,
+      "loss": 0.8559,
+      "step": 6367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.230802705331776e-06,
+      "loss": 0.8355,
+      "step": 6368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.228065919122463e-06,
+      "loss": 0.8904,
+      "step": 6369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225329191076358e-06,
+      "loss": 0.8791,
+      "step": 6370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.222592521399666e-06,
+      "loss": 0.892,
+      "step": 6371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219855910298593e-06,
+      "loss": 0.9594,
+      "step": 6372
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217119357979331e-06,
+      "loss": 0.9185,
+      "step": 6373
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214382864648075e-06,
+      "loss": 0.8384,
+      "step": 6374
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.211646430511008e-06,
+      "loss": 0.8697,
+      "step": 6375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208910055774321e-06,
+      "loss": 0.8819,
+      "step": 6376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206173740644186e-06,
+      "loss": 0.9477,
+      "step": 6377
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203437485326775e-06,
+      "loss": 0.8867,
+      "step": 6378
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.200701290028265e-06,
+      "loss": 0.9066,
+      "step": 6379
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.197965154954815e-06,
+      "loss": 0.9288,
+      "step": 6380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195229080312592e-06,
+      "loss": 0.8761,
+      "step": 6381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192493066307745e-06,
+      "loss": 0.8816,
+      "step": 6382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.189757113146431e-06,
+      "loss": 0.888,
+      "step": 6383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187021221034792e-06,
+      "loss": 0.9133,
+      "step": 6384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184285390178978e-06,
+      "loss": 0.8885,
+      "step": 6385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181549620785115e-06,
+      "loss": 0.8703,
+      "step": 6386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.178813913059348e-06,
+      "loss": 0.8863,
+      "step": 6387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.176078267207799e-06,
+      "loss": 0.8925,
+      "step": 6388
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.173342683436593e-06,
+      "loss": 0.9218,
+      "step": 6389
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170607161951848e-06,
+      "loss": 0.8928,
+      "step": 6390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.167871702959678e-06,
+      "loss": 0.8836,
+      "step": 6391
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.165136306666196e-06,
+      "loss": 0.8562,
+      "step": 6392
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.162400973277503e-06,
+      "loss": 0.8996,
+      "step": 6393
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159665702999705e-06,
+      "loss": 0.5043,
+      "step": 6394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15693049603889e-06,
+      "loss": 0.9165,
+      "step": 6395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.154195352601155e-06,
+      "loss": 0.8828,
+      "step": 6396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.151460272892583e-06,
+      "loss": 0.8489,
+      "step": 6397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148725257119253e-06,
+      "loss": 0.8735,
+      "step": 6398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.145990305487246e-06,
+      "loss": 0.8841,
+      "step": 6399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.143255418202627e-06,
+      "loss": 0.8728,
+      "step": 6400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140520595471473e-06,
+      "loss": 0.8783,
+      "step": 6401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137785837499833e-06,
+      "loss": 0.8794,
+      "step": 6402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.135051144493774e-06,
+      "loss": 0.9111,
+      "step": 6403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.132316516659341e-06,
+      "loss": 0.9003,
+      "step": 6404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129581954202588e-06,
+      "loss": 0.9188,
+      "step": 6405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126847457329547e-06,
+      "loss": 0.8274,
+      "step": 6406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.124113026246268e-06,
+      "loss": 0.8888,
+      "step": 6407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121378661158774e-06,
+      "loss": 0.9084,
+      "step": 6408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118644362273094e-06,
+      "loss": 0.833,
+      "step": 6409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115910129795252e-06,
+      "loss": 0.8832,
+      "step": 6410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113175963931264e-06,
+      "loss": 0.8809,
+      "step": 6411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.110441864887144e-06,
+      "loss": 0.9362,
+      "step": 6412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107707832868896e-06,
+      "loss": 0.912,
+      "step": 6413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.104973868082525e-06,
+      "loss": 0.8806,
+      "step": 6414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102239970734024e-06,
+      "loss": 0.8386,
+      "step": 6415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.099506141029393e-06,
+      "loss": 0.8953,
+      "step": 6416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096772379174608e-06,
+      "loss": 0.8748,
+      "step": 6417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.094038685375663e-06,
+      "loss": 0.9054,
+      "step": 6418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091305059838528e-06,
+      "loss": 0.8839,
+      "step": 6419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08857150276917e-06,
+      "loss": 0.8799,
+      "step": 6420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085838014373564e-06,
+      "loss": 0.8743,
+      "step": 6421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.083104594857663e-06,
+      "loss": 0.5044,
+      "step": 6422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.080371244427435e-06,
+      "loss": 0.9255,
+      "step": 6423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.077637963288814e-06,
+      "loss": 0.9272,
+      "step": 6424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074904751647762e-06,
+      "loss": 0.8898,
+      "step": 6425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.072171609710207e-06,
+      "loss": 0.8866,
+      "step": 6426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069438537682096e-06,
+      "loss": 0.8477,
+      "step": 6427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.066705535769345e-06,
+      "loss": 0.91,
+      "step": 6428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06397260417789e-06,
+      "loss": 0.9048,
+      "step": 6429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061239743113643e-06,
+      "loss": 0.8983,
+      "step": 6430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.058506952782522e-06,
+      "loss": 0.8754,
+      "step": 6431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.055774233390435e-06,
+      "loss": 0.8886,
+      "step": 6432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.053041585143282e-06,
+      "loss": 0.8947,
+      "step": 6433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.050309008246965e-06,
+      "loss": 0.8579,
+      "step": 6434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.047576502907375e-06,
+      "loss": 0.8902,
+      "step": 6435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.044844069330398e-06,
+      "loss": 0.878,
+      "step": 6436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.042111707721916e-06,
+      "loss": 0.9782,
+      "step": 6437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.03937941828781e-06,
+      "loss": 0.5454,
+      "step": 6438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.036647201233947e-06,
+      "loss": 0.9374,
+      "step": 6439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.033915056766187e-06,
+      "loss": 0.9065,
+      "step": 6440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.0311829850904e-06,
+      "loss": 0.85,
+      "step": 6441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.028450986412432e-06,
+      "loss": 0.9221,
+      "step": 6442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.025719060938138e-06,
+      "loss": 0.5101,
+      "step": 6443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022987208873355e-06,
+      "loss": 0.877,
+      "step": 6444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020255430423932e-06,
+      "loss": 0.8562,
+      "step": 6445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.017523725795688e-06,
+      "loss": 0.5939,
+      "step": 6446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.014792095194459e-06,
+      "loss": 0.8737,
+      "step": 6447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01206053882606e-06,
+      "loss": 0.9122,
+      "step": 6448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.009329056896314e-06,
+      "loss": 0.8836,
+      "step": 6449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006597649611026e-06,
+      "loss": 0.9038,
+      "step": 6450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003866317176e-06,
+      "loss": 0.9214,
+      "step": 6451
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.001135059797036e-06,
+      "loss": 0.8811,
+      "step": 6452
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.998403877679926e-06,
+      "loss": 0.9078,
+      "step": 6453
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99567277103046e-06,
+      "loss": 0.9043,
+      "step": 6454
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.992941740054418e-06,
+      "loss": 0.9219,
+      "step": 6455
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.990210784957577e-06,
+      "loss": 0.8442,
+      "step": 6456
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.987479905945704e-06,
+      "loss": 0.8803,
+      "step": 6457
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984749103224568e-06,
+      "loss": 0.9121,
+      "step": 6458
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982018376999924e-06,
+      "loss": 0.8621,
+      "step": 6459
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97928772747753e-06,
+      "loss": 0.8656,
+      "step": 6460
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97655715486313e-06,
+      "loss": 0.8635,
+      "step": 6461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.973826659362464e-06,
+      "loss": 0.8714,
+      "step": 6462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.971096241181272e-06,
+      "loss": 0.8999,
+      "step": 6463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.968365900525275e-06,
+      "loss": 0.874,
+      "step": 6464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96563563760021e-06,
+      "loss": 0.8807,
+      "step": 6465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962905452611785e-06,
+      "loss": 0.9078,
+      "step": 6466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.960175345765718e-06,
+      "loss": 0.8879,
+      "step": 6467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.957445317267708e-06,
+      "loss": 0.8766,
+      "step": 6468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.8641,
+      "step": 6469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.951985496138679e-06,
+      "loss": 0.8385,
+      "step": 6470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94925570391904e-06,
+      "loss": 0.8941,
+      "step": 6471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.946525990870228e-06,
+      "loss": 0.8436,
+      "step": 6472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94379635719792e-06,
+      "loss": 0.8859,
+      "step": 6473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941066803107789e-06,
+      "loss": 0.886,
+      "step": 6474
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9383373288055e-06,
+      "loss": 0.9163,
+      "step": 6475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.935607934496708e-06,
+      "loss": 0.8918,
+      "step": 6476
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93287862038707e-06,
+      "loss": 0.8605,
+      "step": 6477
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.93014938668223e-06,
+      "loss": 0.8724,
+      "step": 6478
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92742023358783e-06,
+      "loss": 0.9016,
+      "step": 6479
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924691161309503e-06,
+      "loss": 0.9093,
+      "step": 6480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.921962170052882e-06,
+      "loss": 0.8962,
+      "step": 6481
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91923326002358e-06,
+      "loss": 0.8778,
+      "step": 6482
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916504431427224e-06,
+      "loss": 0.8464,
+      "step": 6483
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913775684469414e-06,
+      "loss": 0.8891,
+      "step": 6484
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911047019355763e-06,
+      "loss": 0.9313,
+      "step": 6485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908318436291862e-06,
+      "loss": 0.8933,
+      "step": 6486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.905589935483309e-06,
+      "loss": 0.9126,
+      "step": 6487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902861517135682e-06,
+      "loss": 0.9538,
+      "step": 6488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.900133181454566e-06,
+      "loss": 0.9046,
+      "step": 6489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897404928645529e-06,
+      "loss": 0.5044,
+      "step": 6490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.894676758914145e-06,
+      "loss": 0.8579,
+      "step": 6491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89194867246597e-06,
+      "loss": 0.9369,
+      "step": 6492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889220669506554e-06,
+      "loss": 0.8468,
+      "step": 6493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88649275024145e-06,
+      "loss": 0.8816,
+      "step": 6494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8837649148762e-06,
+      "loss": 0.8718,
+      "step": 6495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.881037163616336e-06,
+      "loss": 0.8902,
+      "step": 6496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87830949666739e-06,
+      "loss": 0.871,
+      "step": 6497
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.875581914234883e-06,
+      "loss": 0.9296,
+      "step": 6498
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872854416524332e-06,
+      "loss": 0.9065,
+      "step": 6499
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.870127003741245e-06,
+      "loss": 0.8933,
+      "step": 6500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867399676091127e-06,
+      "loss": 0.8592,
+      "step": 6501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864672433779476e-06,
+      "loss": 0.8473,
+      "step": 6502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861945277011782e-06,
+      "loss": 0.9077,
+      "step": 6503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859218205993524e-06,
+      "loss": 0.9234,
+      "step": 6504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856491220930188e-06,
+      "loss": 0.9236,
+      "step": 6505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.853764322027239e-06,
+      "loss": 0.8548,
+      "step": 6506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.851037509490145e-06,
+      "loss": 0.8779,
+      "step": 6507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84831078352436e-06,
+      "loss": 0.8886,
+      "step": 6508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.845584144335344e-06,
+      "loss": 0.862,
+      "step": 6509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842857592128531e-06,
+      "loss": 0.8986,
+      "step": 6510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84013112710937e-06,
+      "loss": 0.854,
+      "step": 6511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837404749483285e-06,
+      "loss": 0.867,
+      "step": 6512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.834678459455709e-06,
+      "loss": 0.8717,
+      "step": 6513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831952257232056e-06,
+      "loss": 0.8694,
+      "step": 6514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.829226143017735e-06,
+      "loss": 0.8926,
+      "step": 6515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.826500117018157e-06,
+      "loss": 0.8553,
+      "step": 6516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82377417943872e-06,
+      "loss": 0.8894,
+      "step": 6517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.821048330484814e-06,
+      "loss": 0.8903,
+      "step": 6518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.818322570361826e-06,
+      "loss": 0.8868,
+      "step": 6519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.815596899275136e-06,
+      "loss": 0.8744,
+      "step": 6520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.812871317430114e-06,
+      "loss": 0.9035,
+      "step": 6521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810145825032127e-06,
+      "loss": 0.8899,
+      "step": 6522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.807420422286535e-06,
+      "loss": 0.9181,
+      "step": 6523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804695109398686e-06,
+      "loss": 0.9049,
+      "step": 6524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80196988657393e-06,
+      "loss": 0.9031,
+      "step": 6525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799244754017599e-06,
+      "loss": 0.8803,
+      "step": 6526
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796519711935032e-06,
+      "loss": 0.8849,
+      "step": 6527
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.793794760531546e-06,
+      "loss": 0.8723,
+      "step": 6528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.791069900012468e-06,
+      "loss": 0.9259,
+      "step": 6529
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.788345130583099e-06,
+      "loss": 0.8792,
+      "step": 6530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.785620452448753e-06,
+      "loss": 0.9186,
+      "step": 6531
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.782895865814719e-06,
+      "loss": 0.8766,
+      "step": 6532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.780171370886295e-06,
+      "loss": 0.9343,
+      "step": 6533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.777446967868761e-06,
+      "loss": 0.8426,
+      "step": 6534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.77472265696739e-06,
+      "loss": 0.8639,
+      "step": 6535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771998438387458e-06,
+      "loss": 0.9002,
+      "step": 6536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769274312334222e-06,
+      "loss": 0.9107,
+      "step": 6537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76655027901294e-06,
+      "loss": 0.8909,
+      "step": 6538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763826338628865e-06,
+      "loss": 0.8997,
+      "step": 6539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.761102491387232e-06,
+      "loss": 0.9059,
+      "step": 6540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75837873749328e-06,
+      "loss": 0.8518,
+      "step": 6541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755655077152237e-06,
+      "loss": 0.9,
+      "step": 6542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.75293151056932e-06,
+      "loss": 0.9068,
+      "step": 6543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.750208037949747e-06,
+      "loss": 0.8611,
+      "step": 6544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747484659498725e-06,
+      "loss": 0.9092,
+      "step": 6545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.744761375421443e-06,
+      "loss": 0.8871,
+      "step": 6546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.742038185923108e-06,
+      "loss": 0.8991,
+      "step": 6547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739315091208893e-06,
+      "loss": 0.5453,
+      "step": 6548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.736592091483987e-06,
+      "loss": 0.8816,
+      "step": 6549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.73386918695355e-06,
+      "loss": 0.8693,
+      "step": 6550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.731146377822755e-06,
+      "loss": 0.9251,
+      "step": 6551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.728423664296752e-06,
+      "loss": 0.9536,
+      "step": 6552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.725701046580694e-06,
+      "loss": 0.8619,
+      "step": 6553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.72297852487972e-06,
+      "loss": 0.5742,
+      "step": 6554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.720256099398972e-06,
+      "loss": 0.9174,
+      "step": 6555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.71753377034357e-06,
+      "loss": 0.91,
+      "step": 6556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.714811537918634e-06,
+      "loss": 0.9099,
+      "step": 6557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.712089402329285e-06,
+      "loss": 0.8779,
+      "step": 6558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.709367363780618e-06,
+      "loss": 0.9047,
+      "step": 6559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706645422477739e-06,
+      "loss": 0.8606,
+      "step": 6560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703923578625738e-06,
+      "loss": 0.8981,
+      "step": 6561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701201832429698e-06,
+      "loss": 0.9035,
+      "step": 6562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698480184094695e-06,
+      "loss": 0.8635,
+      "step": 6563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.695758633825799e-06,
+      "loss": 0.8904,
+      "step": 6564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693037181828073e-06,
+      "loss": 0.8722,
+      "step": 6565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.690315828306566e-06,
+      "loss": 0.9574,
+      "step": 6566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.687594573466333e-06,
+      "loss": 0.8675,
+      "step": 6567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.684873417512403e-06,
+      "loss": 0.9075,
+      "step": 6568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.682152360649819e-06,
+      "loss": 0.9472,
+      "step": 6569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.679431403083596e-06,
+      "loss": 0.8761,
+      "step": 6570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67671054501876e-06,
+      "loss": 0.8419,
+      "step": 6571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67398978666031e-06,
+      "loss": 0.8919,
+      "step": 6572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67126912821326e-06,
+      "loss": 0.8904,
+      "step": 6573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.668548569882593e-06,
+      "loss": 0.871,
+      "step": 6574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.665828111873307e-06,
+      "loss": 0.8803,
+      "step": 6575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663107754390375e-06,
+      "loss": 0.8827,
+      "step": 6576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.660387497638765e-06,
+      "loss": 0.8444,
+      "step": 6577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.657667341823449e-06,
+      "loss": 0.9079,
+      "step": 6578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654947287149376e-06,
+      "loss": 0.9043,
+      "step": 6579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652227333821504e-06,
+      "loss": 0.8907,
+      "step": 6580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.649507482044766e-06,
+      "loss": 0.9008,
+      "step": 6581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6467877320241e-06,
+      "loss": 0.8963,
+      "step": 6582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644068083964435e-06,
+      "loss": 0.8965,
+      "step": 6583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641348538070683e-06,
+      "loss": 0.8832,
+      "step": 6584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638629094547757e-06,
+      "loss": 0.8893,
+      "step": 6585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63590975360056e-06,
+      "loss": 0.8659,
+      "step": 6586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.633190515433992e-06,
+      "loss": 0.8993,
+      "step": 6587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63047138025293e-06,
+      "loss": 0.8455,
+      "step": 6588
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627752348262263e-06,
+      "loss": 0.8771,
+      "step": 6589
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.625033419666856e-06,
+      "loss": 0.9008,
+      "step": 6590
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.622314594671582e-06,
+      "loss": 0.8821,
+      "step": 6591
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.619595873481288e-06,
+      "loss": 0.8715,
+      "step": 6592
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616877256300829e-06,
+      "loss": 0.8464,
+      "step": 6593
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.614158743335042e-06,
+      "loss": 0.9187,
+      "step": 6594
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611440334788762e-06,
+      "loss": 0.8234,
+      "step": 6595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.608722030866812e-06,
+      "loss": 0.8541,
+      "step": 6596
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.606003831774013e-06,
+      "loss": 0.9079,
+      "step": 6597
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.603285737715172e-06,
+      "loss": 0.9085,
+      "step": 6598
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600567748895083e-06,
+      "loss": 0.8283,
+      "step": 6599
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597849865518552e-06,
+      "loss": 0.906,
+      "step": 6600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595132087790355e-06,
+      "loss": 0.9181,
+      "step": 6601
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592414415915275e-06,
+      "loss": 0.873,
+      "step": 6602
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589696850098077e-06,
+      "loss": 0.8527,
+      "step": 6603
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.586979390543526e-06,
+      "loss": 0.5141,
+      "step": 6604
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.584262037456374e-06,
+      "loss": 0.9024,
+      "step": 6605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.581544791041367e-06,
+      "loss": 0.8763,
+      "step": 6606
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.578827651503244e-06,
+      "loss": 0.9083,
+      "step": 6607
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576110619046726e-06,
+      "loss": 0.8661,
+      "step": 6608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.573393693876547e-06,
+      "loss": 0.8867,
+      "step": 6609
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570676876197408e-06,
+      "loss": 0.8881,
+      "step": 6610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567960166214026e-06,
+      "loss": 0.9214,
+      "step": 6611
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565243564131087e-06,
+      "loss": 0.8796,
+      "step": 6612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.562527070153288e-06,
+      "loss": 0.8813,
+      "step": 6613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.559810684485302e-06,
+      "loss": 0.9224,
+      "step": 6614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55709440733181e-06,
+      "loss": 0.9554,
+      "step": 6615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55437823889747e-06,
+      "loss": 0.4864,
+      "step": 6616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.551662179386944e-06,
+      "loss": 0.8925,
+      "step": 6617
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.548946229004876e-06,
+      "loss": 0.915,
+      "step": 6618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.546230387955901e-06,
+      "loss": 0.9668,
+      "step": 6619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54351465644466e-06,
+      "loss": 0.8658,
+      "step": 6620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540799034675769e-06,
+      "loss": 0.8215,
+      "step": 6621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.538083522853849e-06,
+      "loss": 0.9154,
+      "step": 6622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5353681211835e-06,
+      "loss": 0.8973,
+      "step": 6623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53265282986933e-06,
+      "loss": 0.89,
+      "step": 6624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.529937649115916e-06,
+      "loss": 0.9009,
+      "step": 6625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52722257912785e-06,
+      "loss": 0.9332,
+      "step": 6626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5245076201097e-06,
+      "loss": 0.8875,
+      "step": 6627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521792772266037e-06,
+      "loss": 0.8963,
+      "step": 6628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.519078035801415e-06,
+      "loss": 0.9148,
+      "step": 6629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516363410920376e-06,
+      "loss": 0.8792,
+      "step": 6630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.51364889782747e-06,
+      "loss": 0.9399,
+      "step": 6631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.510934496727218e-06,
+      "loss": 0.894,
+      "step": 6632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.508220207824156e-06,
+      "loss": 0.9021,
+      "step": 6633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.505506031322784e-06,
+      "loss": 0.8335,
+      "step": 6634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.50279196742762e-06,
+      "loss": 0.8745,
+      "step": 6635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500078016343152e-06,
+      "loss": 0.8814,
+      "step": 6636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497364178273878e-06,
+      "loss": 0.8568,
+      "step": 6637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.494650453424272e-06,
+      "loss": 0.9085,
+      "step": 6638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491936841998811e-06,
+      "loss": 0.8281,
+      "step": 6639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489223344201958e-06,
+      "loss": 0.8507,
+      "step": 6640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48650996023816e-06,
+      "loss": 0.8892,
+      "step": 6641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.483796690311875e-06,
+      "loss": 0.8735,
+      "step": 6642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48108353462753e-06,
+      "loss": 0.8802,
+      "step": 6643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478370493389563e-06,
+      "loss": 0.871,
+      "step": 6644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.475657566802388e-06,
+      "loss": 0.9181,
+      "step": 6645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472944755070423e-06,
+      "loss": 0.8557,
+      "step": 6646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470232058398063e-06,
+      "loss": 0.8718,
+      "step": 6647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467519476989711e-06,
+      "loss": 0.8438,
+      "step": 6648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.464807011049751e-06,
+      "loss": 0.8709,
+      "step": 6649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.462094660782555e-06,
+      "loss": 0.8363,
+      "step": 6650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4593824263925e-06,
+      "loss": 0.8976,
+      "step": 6651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.456670308083936e-06,
+      "loss": 0.8302,
+      "step": 6652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.453958306061223e-06,
+      "loss": 0.896,
+      "step": 6653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451246420528694e-06,
+      "loss": 0.8809,
+      "step": 6654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448534651690694e-06,
+      "loss": 0.9032,
+      "step": 6655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.445822999751536e-06,
+      "loss": 0.9119,
+      "step": 6656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443111464915545e-06,
+      "loss": 0.9089,
+      "step": 6657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.440400047387019e-06,
+      "loss": 0.9316,
+      "step": 6658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437688747370267e-06,
+      "loss": 0.8822,
+      "step": 6659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.434977565069572e-06,
+      "loss": 0.8138,
+      "step": 6660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.432266500689213e-06,
+      "loss": 0.8818,
+      "step": 6661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.429555554433466e-06,
+      "loss": 0.8325,
+      "step": 6662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.426844726506589e-06,
+      "loss": 0.5193,
+      "step": 6663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.424134017112841e-06,
+      "loss": 0.9146,
+      "step": 6664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42142342645646e-06,
+      "loss": 0.887,
+      "step": 6665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418712954741691e-06,
+      "loss": 0.8599,
+      "step": 6666
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.416002602172752e-06,
+      "loss": 0.8824,
+      "step": 6667
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.41329236895387e-06,
+      "loss": 0.8664,
+      "step": 6668
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.410582255289243e-06,
+      "loss": 0.9255,
+      "step": 6669
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.40787226138308e-06,
+      "loss": 0.8461,
+      "step": 6670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.40516238743957e-06,
+      "loss": 0.9008,
+      "step": 6671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.40245263366289e-06,
+      "loss": 0.8571,
+      "step": 6672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.39974300025722e-06,
+      "loss": 0.8953,
+      "step": 6673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.397033487426717e-06,
+      "loss": 0.8685,
+      "step": 6674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.394324095375543e-06,
+      "loss": 0.872,
+      "step": 6675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.391614824307835e-06,
+      "loss": 0.8342,
+      "step": 6676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38890567442774e-06,
+      "loss": 0.8646,
+      "step": 6677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.386196645939372e-06,
+      "loss": 0.8819,
+      "step": 6678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383487739046865e-06,
+      "loss": 0.925,
+      "step": 6679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.380778953954314e-06,
+      "loss": 0.9201,
+      "step": 6680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.378070290865827e-06,
+      "loss": 0.8711,
+      "step": 6681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.375361749985494e-06,
+      "loss": 0.8474,
+      "step": 6682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37265333151739e-06,
+      "loss": 0.8468,
+      "step": 6683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369945035665595e-06,
+      "loss": 0.4769,
+      "step": 6684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.367236862634166e-06,
+      "loss": 0.9038,
+      "step": 6685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364528812627164e-06,
+      "loss": 0.872,
+      "step": 6686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.361820885848623e-06,
+      "loss": 0.8485,
+      "step": 6687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35911308250259e-06,
+      "loss": 0.897,
+      "step": 6688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356405402793081e-06,
+      "loss": 0.9627,
+      "step": 6689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35369784692412e-06,
+      "loss": 0.9259,
+      "step": 6690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35099041509971e-06,
+      "loss": 0.9307,
+      "step": 6691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348283107523848e-06,
+      "loss": 0.6502,
+      "step": 6692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345575924400527e-06,
+      "loss": 0.9107,
+      "step": 6693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34286886593372e-06,
+      "loss": 0.893,
+      "step": 6694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340161932327405e-06,
+      "loss": 0.4886,
+      "step": 6695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337455123785533e-06,
+      "loss": 0.8492,
+      "step": 6696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.334748440512064e-06,
+      "loss": 0.8953,
+      "step": 6697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33204188271093e-06,
+      "loss": 0.9269,
+      "step": 6698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329335450586075e-06,
+      "loss": 0.8891,
+      "step": 6699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.326629144341408e-06,
+      "loss": 0.8535,
+      "step": 6700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.323922964180853e-06,
+      "loss": 0.9508,
+      "step": 6701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.321216910308309e-06,
+      "loss": 0.872,
+      "step": 6702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318510982927668e-06,
+      "loss": 0.5181,
+      "step": 6703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31580518224282e-06,
+      "loss": 0.8648,
+      "step": 6704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.313099508457633e-06,
+      "loss": 0.8729,
+      "step": 6705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31039396177598e-06,
+      "loss": 0.8666,
+      "step": 6706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.307688542401709e-06,
+      "loss": 0.9594,
+      "step": 6707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.304983250538676e-06,
+      "loss": 0.8433,
+      "step": 6708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.302278086390706e-06,
+      "loss": 0.9098,
+      "step": 6709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299573050161637e-06,
+      "loss": 0.9025,
+      "step": 6710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.296868142055278e-06,
+      "loss": 0.9187,
+      "step": 6711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294163362275444e-06,
+      "loss": 0.8992,
+      "step": 6712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.291458711025928e-06,
+      "loss": 0.8972,
+      "step": 6713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.288754188510516e-06,
+      "loss": 0.8807,
+      "step": 6714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286049794932997e-06,
+      "loss": 0.9012,
+      "step": 6715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.283345530497128e-06,
+      "loss": 0.8877,
+      "step": 6716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280641395406676e-06,
+      "loss": 0.9099,
+      "step": 6717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277937389865386e-06,
+      "loss": 0.9058,
+      "step": 6718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.275233514077003e-06,
+      "loss": 0.9194,
+      "step": 6719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272529768245251e-06,
+      "loss": 0.8862,
+      "step": 6720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269826152573857e-06,
+      "loss": 0.8736,
+      "step": 6721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26712266726652e-06,
+      "loss": 0.9174,
+      "step": 6722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264419312526955e-06,
+      "loss": 0.8831,
+      "step": 6723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.261716088558846e-06,
+      "loss": 0.9388,
+      "step": 6724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.259012995565868e-06,
+      "loss": 0.893,
+      "step": 6725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.2563100337517e-06,
+      "loss": 0.522,
+      "step": 6726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253607203319997e-06,
+      "loss": 0.8156,
+      "step": 6727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25090450447442e-06,
+      "loss": 0.8566,
+      "step": 6728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.248201937418598e-06,
+      "loss": 0.918,
+      "step": 6729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245499502356172e-06,
+      "loss": 0.893,
+      "step": 6730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.242797199490757e-06,
+      "loss": 0.8465,
+      "step": 6731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24009502902597e-06,
+      "loss": 0.9246,
+      "step": 6732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.237392991165407e-06,
+      "loss": 0.9031,
+      "step": 6733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234691086112662e-06,
+      "loss": 0.832,
+      "step": 6734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.231989314071318e-06,
+      "loss": 0.8708,
+      "step": 6735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229287675244942e-06,
+      "loss": 0.9021,
+      "step": 6736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226586169837103e-06,
+      "loss": 0.9191,
+      "step": 6737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.223884798051341e-06,
+      "loss": 0.8644,
+      "step": 6738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221183560091211e-06,
+      "loss": 0.9025,
+      "step": 6739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21848245616023e-06,
+      "loss": 0.914,
+      "step": 6740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215781486461933e-06,
+      "loss": 0.9181,
+      "step": 6741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213080651199819e-06,
+      "loss": 0.9163,
+      "step": 6742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210379950577398e-06,
+      "loss": 0.8119,
+      "step": 6743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.207679384798159e-06,
+      "loss": 0.9151,
+      "step": 6744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204978954065574e-06,
+      "loss": 0.8468,
+      "step": 6745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202278658583125e-06,
+      "loss": 0.9357,
+      "step": 6746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.199578498554262e-06,
+      "loss": 0.9231,
+      "step": 6747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.196878474182443e-06,
+      "loss": 0.8673,
+      "step": 6748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194178585671102e-06,
+      "loss": 0.8535,
+      "step": 6749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191478833223675e-06,
+      "loss": 0.9282,
+      "step": 6750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.188779217043573e-06,
+      "loss": 0.9023,
+      "step": 6751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186079737334214e-06,
+      "loss": 0.912,
+      "step": 6752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183380394298987e-06,
+      "loss": 0.9063,
+      "step": 6753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180681188141292e-06,
+      "loss": 0.8903,
+      "step": 6754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.177982119064497e-06,
+      "loss": 0.8636,
+      "step": 6755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.175283187271974e-06,
+      "loss": 0.8589,
+      "step": 6756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172584392967078e-06,
+      "loss": 0.9197,
+      "step": 6757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.169885736353158e-06,
+      "loss": 0.8761,
+      "step": 6758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.167187217633555e-06,
+      "loss": 0.8773,
+      "step": 6759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164488837011585e-06,
+      "loss": 0.9102,
+      "step": 6760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161790594690577e-06,
+      "loss": 0.8945,
+      "step": 6761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.159092490873824e-06,
+      "loss": 0.8552,
+      "step": 6762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15639452576463e-06,
+      "loss": 0.889,
+      "step": 6763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153696699566272e-06,
+      "loss": 0.9375,
+      "step": 6764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150999012482031e-06,
+      "loss": 0.8537,
+      "step": 6765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.14830146471517e-06,
+      "loss": 0.8763,
+      "step": 6766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.145604056468935e-06,
+      "loss": 0.8975,
+      "step": 6767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142906787946578e-06,
+      "loss": 0.8958,
+      "step": 6768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140209659351322e-06,
+      "loss": 0.8762,
+      "step": 6769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.137512670886397e-06,
+      "loss": 0.8852,
+      "step": 6770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134815822755008e-06,
+      "loss": 0.893,
+      "step": 6771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13211911516036e-06,
+      "loss": 0.911,
+      "step": 6772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.129422548305637e-06,
+      "loss": 0.8774,
+      "step": 6773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126726122394027e-06,
+      "loss": 0.8893,
+      "step": 6774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.124029837628691e-06,
+      "loss": 0.8588,
+      "step": 6775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12133369421279e-06,
+      "loss": 0.8858,
+      "step": 6776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118637692349468e-06,
+      "loss": 0.8813,
+      "step": 6777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115941832241867e-06,
+      "loss": 0.8366,
+      "step": 6778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11324611409311e-06,
+      "loss": 0.895,
+      "step": 6779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11055053810631e-06,
+      "loss": 0.9122,
+      "step": 6780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10785510448458e-06,
+      "loss": 0.8748,
+      "step": 6781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.105159813431002e-06,
+      "loss": 0.8632,
+      "step": 6782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10246466514867e-06,
+      "loss": 0.8983,
+      "step": 6783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.099769659840648e-06,
+      "loss": 0.8664,
+      "step": 6784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.097074797710007e-06,
+      "loss": 0.8485,
+      "step": 6785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.09438007895979e-06,
+      "loss": 0.8774,
+      "step": 6786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.091685503793035e-06,
+      "loss": 0.5633,
+      "step": 6787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.08899107241278e-06,
+      "loss": 0.8586,
+      "step": 6788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.086296785022034e-06,
+      "loss": 0.8634,
+      "step": 6789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.083602641823814e-06,
+      "loss": 0.8885,
+      "step": 6790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.080908643021107e-06,
+      "loss": 0.8835,
+      "step": 6791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.078214788816906e-06,
+      "loss": 0.8691,
+      "step": 6792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.075521079414182e-06,
+      "loss": 0.8751,
+      "step": 6793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.072827515015902e-06,
+      "loss": 0.8834,
+      "step": 6794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.070134095825016e-06,
+      "loss": 0.8611,
+      "step": 6795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06744082204447e-06,
+      "loss": 0.949,
+      "step": 6796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06474769387719e-06,
+      "loss": 0.9072,
+      "step": 6797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062054711526098e-06,
+      "loss": 0.8964,
+      "step": 6798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.059361875194105e-06,
+      "loss": 0.8945,
+      "step": 6799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.056669185084108e-06,
+      "loss": 0.8851,
+      "step": 6800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053976641398993e-06,
+      "loss": 0.9194,
+      "step": 6801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051284244341636e-06,
+      "loss": 0.8318,
+      "step": 6802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.048591994114906e-06,
+      "loss": 0.8937,
+      "step": 6803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045899890921652e-06,
+      "loss": 0.8984,
+      "step": 6804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.043207934964722e-06,
+      "loss": 0.8679,
+      "step": 6805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04051612644694e-06,
+      "loss": 0.8607,
+      "step": 6806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.037824465571138e-06,
+      "loss": 0.9235,
+      "step": 6807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035132952540118e-06,
+      "loss": 0.5445,
+      "step": 6808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.032441587556676e-06,
+      "loss": 0.8976,
+      "step": 6809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.029750370823606e-06,
+      "loss": 0.8892,
+      "step": 6810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.027059302543678e-06,
+      "loss": 0.8568,
+      "step": 6811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.024368382919665e-06,
+      "loss": 0.8673,
+      "step": 6812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.021677612154311e-06,
+      "loss": 0.8807,
+      "step": 6813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018986990450368e-06,
+      "loss": 0.8867,
+      "step": 6814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.016296518010558e-06,
+      "loss": 0.8607,
+      "step": 6815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01360619503761e-06,
+      "loss": 0.8374,
+      "step": 6816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010916021734225e-06,
+      "loss": 0.9095,
+      "step": 6817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008225998303107e-06,
+      "loss": 0.913,
+      "step": 6818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005536124946936e-06,
+      "loss": 0.8323,
+      "step": 6819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002846401868392e-06,
+      "loss": 0.9046,
+      "step": 6820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.000156829270136e-06,
+      "loss": 0.8745,
+      "step": 6821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997467407354821e-06,
+      "loss": 0.9133,
+      "step": 6822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994778136325089e-06,
+      "loss": 0.8815,
+      "step": 6823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992089016383565e-06,
+      "loss": 0.8675,
+      "step": 6824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989400047732875e-06,
+      "loss": 0.8773,
+      "step": 6825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986711230575617e-06,
+      "loss": 0.8732,
+      "step": 6826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984022565114396e-06,
+      "loss": 0.8969,
+      "step": 6827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981334051551788e-06,
+      "loss": 0.9064,
+      "step": 6828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978645690090365e-06,
+      "loss": 0.8895,
+      "step": 6829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.975957480932695e-06,
+      "loss": 0.9014,
+      "step": 6830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97326942428132e-06,
+      "loss": 0.8981,
+      "step": 6831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.970581520338785e-06,
+      "loss": 0.818,
+      "step": 6832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967893769307608e-06,
+      "loss": 0.879,
+      "step": 6833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.965206171390315e-06,
+      "loss": 0.8692,
+      "step": 6834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.962518726789397e-06,
+      "loss": 0.8942,
+      "step": 6835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.959831435707357e-06,
+      "loss": 0.8718,
+      "step": 6836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957144298346667e-06,
+      "loss": 0.915,
+      "step": 6837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954457314909805e-06,
+      "loss": 0.8862,
+      "step": 6838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951770485599218e-06,
+      "loss": 0.9123,
+      "step": 6839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.949083810617358e-06,
+      "loss": 0.8695,
+      "step": 6840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946397290166655e-06,
+      "loss": 0.8514,
+      "step": 6841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.943710924449535e-06,
+      "loss": 0.8747,
+      "step": 6842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941024713668406e-06,
+      "loss": 0.882,
+      "step": 6843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938338658025669e-06,
+      "loss": 0.9127,
+      "step": 6844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935652757723709e-06,
+      "loss": 0.8247,
+      "step": 6845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932967012964901e-06,
+      "loss": 0.8692,
+      "step": 6846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.930281423951615e-06,
+      "loss": 0.9225,
+      "step": 6847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927595990886194e-06,
+      "loss": 0.8959,
+      "step": 6848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.924910713970986e-06,
+      "loss": 0.8876,
+      "step": 6849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922225593408318e-06,
+      "loss": 0.8625,
+      "step": 6850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9195406294005e-06,
+      "loss": 0.8288,
+      "step": 6851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.916855822149847e-06,
+      "loss": 0.8066,
+      "step": 6852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914171171858645e-06,
+      "loss": 0.8565,
+      "step": 6853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91148667872918e-06,
+      "loss": 0.8799,
+      "step": 6854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908802342963715e-06,
+      "loss": 0.8579,
+      "step": 6855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.906118164764518e-06,
+      "loss": 0.9065,
+      "step": 6856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903434144333824e-06,
+      "loss": 0.8953,
+      "step": 6857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.900750281873875e-06,
+      "loss": 0.9069,
+      "step": 6858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89806657758689e-06,
+      "loss": 0.9001,
+      "step": 6859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895383031675074e-06,
+      "loss": 0.9283,
+      "step": 6860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892699644340633e-06,
+      "loss": 0.8995,
+      "step": 6861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89001641578575e-06,
+      "loss": 0.8886,
+      "step": 6862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8873333462126e-06,
+      "loss": 0.8752,
+      "step": 6863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884650435823345e-06,
+      "loss": 0.8791,
+      "step": 6864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881967684820134e-06,
+      "loss": 0.9316,
+      "step": 6865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.879285093405105e-06,
+      "loss": 0.912,
+      "step": 6866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876602661780386e-06,
+      "loss": 0.8724,
+      "step": 6867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873920390148089e-06,
+      "loss": 0.861,
+      "step": 6868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871238278710322e-06,
+      "loss": 0.8284,
+      "step": 6869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868556327669172e-06,
+      "loss": 0.9218,
+      "step": 6870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865874537226709e-06,
+      "loss": 0.8714,
+      "step": 6871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.863192907585013e-06,
+      "loss": 0.8821,
+      "step": 6872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.860511438946123e-06,
+      "loss": 0.8603,
+      "step": 6873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.857830131512094e-06,
+      "loss": 0.8484,
+      "step": 6874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.855148985484946e-06,
+      "loss": 0.9521,
+      "step": 6875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.852468001066703e-06,
+      "loss": 0.8559,
+      "step": 6876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.849787178459363e-06,
+      "loss": 0.9254,
+      "step": 6877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847106517864927e-06,
+      "loss": 0.8666,
+      "step": 6878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.844426019485368e-06,
+      "loss": 0.8967,
+      "step": 6879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841745683522662e-06,
+      "loss": 0.9151,
+      "step": 6880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.839065510178763e-06,
+      "loss": 0.8457,
+      "step": 6881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836385499655607e-06,
+      "loss": 0.847,
+      "step": 6882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.833705652155136e-06,
+      "loss": 0.8427,
+      "step": 6883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.831025967879265e-06,
+      "loss": 0.9019,
+      "step": 6884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.828346447029903e-06,
+      "loss": 0.8856,
+      "step": 6885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.825667089808943e-06,
+      "loss": 0.8809,
+      "step": 6886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822987896418269e-06,
+      "loss": 0.8553,
+      "step": 6887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.820308867059751e-06,
+      "loss": 0.9009,
+      "step": 6888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817630001935247e-06,
+      "loss": 0.9098,
+      "step": 6889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814951301246597e-06,
+      "loss": 0.886,
+      "step": 6890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.812272765195646e-06,
+      "loss": 0.8735,
+      "step": 6891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.809594393984207e-06,
+      "loss": 0.8794,
+      "step": 6892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.806916187814084e-06,
+      "loss": 0.8875,
+      "step": 6893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.804238146887081e-06,
+      "loss": 0.9153,
+      "step": 6894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.801560271404976e-06,
+      "loss": 0.9078,
+      "step": 6895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798882561569546e-06,
+      "loss": 0.8918,
+      "step": 6896
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796205017582541e-06,
+      "loss": 0.9589,
+      "step": 6897
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.793527639645715e-06,
+      "loss": 0.9059,
+      "step": 6898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790850427960795e-06,
+      "loss": 0.8232,
+      "step": 6899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.788173382729507e-06,
+      "loss": 0.9151,
+      "step": 6900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.785496504153557e-06,
+      "loss": 0.8986,
+      "step": 6901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.782819792434638e-06,
+      "loss": 0.8922,
+      "step": 6902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.78014324777444e-06,
+      "loss": 0.8655,
+      "step": 6903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.777466870374624e-06,
+      "loss": 0.8656,
+      "step": 6904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.9239,
+      "step": 6905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.772114618162781e-06,
+      "loss": 0.9129,
+      "step": 6906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.769438743754028e-06,
+      "loss": 0.8969,
+      "step": 6907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.766763037412219e-06,
+      "loss": 0.8959,
+      "step": 6908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.764087499338962e-06,
+      "loss": 0.8998,
+      "step": 6909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.761412129735853e-06,
+      "loss": 0.9163,
+      "step": 6910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.758736928804469e-06,
+      "loss": 0.9052,
+      "step": 6911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.756061896746387e-06,
+      "loss": 0.8933,
+      "step": 6912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.753387033763154e-06,
+      "loss": 0.8743,
+      "step": 6913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.750712340056323e-06,
+      "loss": 0.8784,
+      "step": 6914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748037815827417e-06,
+      "loss": 0.944,
+      "step": 6915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745363461277964e-06,
+      "loss": 0.9195,
+      "step": 6916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742689276609459e-06,
+      "loss": 0.9149,
+      "step": 6917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740015262023406e-06,
+      "loss": 0.9036,
+      "step": 6918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737341417721274e-06,
+      "loss": 0.852,
+      "step": 6919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73466774390454e-06,
+      "loss": 0.9202,
+      "step": 6920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.731994240774648e-06,
+      "loss": 0.8627,
+      "step": 6921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72932090853305e-06,
+      "loss": 0.9132,
+      "step": 6922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726647747381171e-06,
+      "loss": 0.915,
+      "step": 6923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.723974757520421e-06,
+      "loss": 0.8954,
+      "step": 6924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72130193915221e-06,
+      "loss": 0.8045,
+      "step": 6925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71862929247792e-06,
+      "loss": 0.8721,
+      "step": 6926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.715956817698938e-06,
+      "loss": 0.9459,
+      "step": 6927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71328451501662e-06,
+      "loss": 0.898,
+      "step": 6928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.710612384632321e-06,
+      "loss": 0.8787,
+      "step": 6929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.707940426747378e-06,
+      "loss": 0.8896,
+      "step": 6930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.705268641563115e-06,
+      "loss": 0.5336,
+      "step": 6931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702597029280848e-06,
+      "loss": 0.8868,
+      "step": 6932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.699925590101868e-06,
+      "loss": 0.9029,
+      "step": 6933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.697254324227472e-06,
+      "loss": 0.9112,
+      "step": 6934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694583231858921e-06,
+      "loss": 0.8916,
+      "step": 6935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.691912313197484e-06,
+      "loss": 0.9,
+      "step": 6936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689241568444402e-06,
+      "loss": 0.874,
+      "step": 6937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.686570997800914e-06,
+      "loss": 0.8862,
+      "step": 6938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.683900601468232e-06,
+      "loss": 0.8879,
+      "step": 6939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681230379647576e-06,
+      "loss": 0.8872,
+      "step": 6940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.678560332540126e-06,
+      "loss": 0.8549,
+      "step": 6941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.675890460347075e-06,
+      "loss": 0.8696,
+      "step": 6942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.673220763269586e-06,
+      "loss": 0.9088,
+      "step": 6943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670551241508809e-06,
+      "loss": 0.8826,
+      "step": 6944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.667881895265895e-06,
+      "loss": 0.8449,
+      "step": 6945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.665212724741961e-06,
+      "loss": 0.8825,
+      "step": 6946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662543730138136e-06,
+      "loss": 0.862,
+      "step": 6947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659874911655506e-06,
+      "loss": 0.8928,
+      "step": 6948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.657206269495172e-06,
+      "loss": 0.9237,
+      "step": 6949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.654537803858205e-06,
+      "loss": 0.8759,
+      "step": 6950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651869514945666e-06,
+      "loss": 0.8809,
+      "step": 6951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.649201402958603e-06,
+      "loss": 0.8815,
+      "step": 6952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646533468098054e-06,
+      "loss": 0.8945,
+      "step": 6953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64386571056504e-06,
+      "loss": 0.8668,
+      "step": 6954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.641198130560566e-06,
+      "loss": 0.9125,
+      "step": 6955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.638530728285633e-06,
+      "loss": 0.8542,
+      "step": 6956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635863503941217e-06,
+      "loss": 0.9228,
+      "step": 6957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.633196457728291e-06,
+      "loss": 0.8861,
+      "step": 6958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.630529589847807e-06,
+      "loss": 0.877,
+      "step": 6959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.627862900500709e-06,
+      "loss": 0.9052,
+      "step": 6960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.62519638988792e-06,
+      "loss": 0.8844,
+      "step": 6961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.622530058210363e-06,
+      "loss": 0.5227,
+      "step": 6962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619863905668929e-06,
+      "loss": 0.925,
+      "step": 6963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.617197932464517e-06,
+      "loss": 0.8794,
+      "step": 6964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.614532138797994e-06,
+      "loss": 0.8919,
+      "step": 6965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611866524870217e-06,
+      "loss": 0.9462,
+      "step": 6966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.609201090882044e-06,
+      "loss": 0.9049,
+      "step": 6967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.606535837034295e-06,
+      "loss": 0.8764,
+      "step": 6968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6038707635278016e-06,
+      "loss": 0.9203,
+      "step": 6969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.601205870563363e-06,
+      "loss": 0.846,
+      "step": 6970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598541158341774e-06,
+      "loss": 0.9189,
+      "step": 6971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.595876627063816e-06,
+      "loss": 0.9495,
+      "step": 6972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.593212276930253e-06,
+      "loss": 0.8884,
+      "step": 6973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5905481081418365e-06,
+      "loss": 0.8603,
+      "step": 6974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5878841208993015e-06,
+      "loss": 0.8911,
+      "step": 6975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585220315403379e-06,
+      "loss": 0.8966,
+      "step": 6976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582556691854772e-06,
+      "loss": 0.9217,
+      "step": 6977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579893250454186e-06,
+      "loss": 0.8467,
+      "step": 6978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.577229991402295e-06,
+      "loss": 0.8927,
+      "step": 6979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.574566914899779e-06,
+      "loss": 0.9066,
+      "step": 6980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.571904021147284e-06,
+      "loss": 0.8905,
+      "step": 6981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5692413103454594e-06,
+      "loss": 0.8503,
+      "step": 6982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.566578782694928e-06,
+      "loss": 0.9048,
+      "step": 6983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.56391643839631e-06,
+      "loss": 0.8947,
+      "step": 6984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.561254277650204e-06,
+      "loss": 0.8699,
+      "step": 6985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55859230065719e-06,
+      "loss": 0.9319,
+      "step": 6986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.555930507617853e-06,
+      "loss": 0.906,
+      "step": 6987
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.553268898732741e-06,
+      "loss": 0.8972,
+      "step": 6988
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.550607474202407e-06,
+      "loss": 0.9192,
+      "step": 6989
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547946234227377e-06,
+      "loss": 0.8752,
+      "step": 6990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545285179008175e-06,
+      "loss": 0.9212,
+      "step": 6991
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.542624308745294e-06,
+      "loss": 0.9106,
+      "step": 6992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.539963623639235e-06,
+      "loss": 0.5051,
+      "step": 6993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.537303123890467e-06,
+      "loss": 0.8784,
+      "step": 6994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534642809699455e-06,
+      "loss": 0.82,
+      "step": 6995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.531982681266645e-06,
+      "loss": 0.8828,
+      "step": 6996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.529322738792468e-06,
+      "loss": 0.9055,
+      "step": 6997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5266629824773506e-06,
+      "loss": 0.8643,
+      "step": 6998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.524003412521691e-06,
+      "loss": 0.9327,
+      "step": 6999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5213440291258876e-06,
+      "loss": 0.9101,
+      "step": 7000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.518684832490311e-06,
+      "loss": 0.9154,
+      "step": 7001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.516025822815331e-06,
+      "loss": 0.8769,
+      "step": 7002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51336700030129e-06,
+      "loss": 0.9098,
+      "step": 7003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510708365148534e-06,
+      "loss": 0.8941,
+      "step": 7004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.508049917557372e-06,
+      "loss": 0.9226,
+      "step": 7005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.50539165772812e-06,
+      "loss": 0.9055,
+      "step": 7006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.50273358586107e-06,
+      "loss": 0.8931,
+      "step": 7007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.500075702156493e-06,
+      "loss": 0.8546,
+      "step": 7008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497418006814664e-06,
+      "loss": 0.8586,
+      "step": 7009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494760500035824e-06,
+      "loss": 0.9121,
+      "step": 7010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4921031820202175e-06,
+      "loss": 0.9145,
+      "step": 7011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489446052968059e-06,
+      "loss": 0.9319,
+      "step": 7012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4867891130795625e-06,
+      "loss": 0.9137,
+      "step": 7013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484132362554915e-06,
+      "loss": 0.9158,
+      "step": 7014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481475801594302e-06,
+      "loss": 0.8822,
+      "step": 7015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478819430397888e-06,
+      "loss": 0.5841,
+      "step": 7016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.476163249165817e-06,
+      "loss": 0.9208,
+      "step": 7017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473507258098235e-06,
+      "loss": 0.8606,
+      "step": 7018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470851457395254e-06,
+      "loss": 0.8421,
+      "step": 7019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46819584725699e-06,
+      "loss": 0.8859,
+      "step": 7020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.465540427883528e-06,
+      "loss": 0.8671,
+      "step": 7021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.462885199474956e-06,
+      "loss": 0.9281,
+      "step": 7022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.46023016223133e-06,
+      "loss": 0.8157,
+      "step": 7023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.457575316352707e-06,
+      "loss": 0.8561,
+      "step": 7024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.454920662039118e-06,
+      "loss": 0.8884,
+      "step": 7025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.452266199490588e-06,
+      "loss": 0.885,
+      "step": 7026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.449611928907123e-06,
+      "loss": 0.8178,
+      "step": 7027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4469578504887094e-06,
+      "loss": 0.8709,
+      "step": 7028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.444303964435333e-06,
+      "loss": 0.8376,
+      "step": 7029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.441650270946952e-06,
+      "loss": 0.883,
+      "step": 7030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43899677022352e-06,
+      "loss": 0.8963,
+      "step": 7031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.436343462464963e-06,
+      "loss": 0.9103,
+      "step": 7032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.433690347871213e-06,
+      "loss": 0.9166,
+      "step": 7033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4310374266421625e-06,
+      "loss": 0.5351,
+      "step": 7034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428384698977713e-06,
+      "loss": 0.8659,
+      "step": 7035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.425732165077733e-06,
+      "loss": 0.8829,
+      "step": 7036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4230798251420865e-06,
+      "loss": 0.8974,
+      "step": 7037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.420427679370623e-06,
+      "loss": 0.5814,
+      "step": 7038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.417775727963169e-06,
+      "loss": 0.8837,
+      "step": 7039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415123971119549e-06,
+      "loss": 0.855,
+      "step": 7040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.41247240903956e-06,
+      "loss": 0.8345,
+      "step": 7041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.409821041922993e-06,
+      "loss": 0.852,
+      "step": 7042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.40716986996962e-06,
+      "loss": 0.9054,
+      "step": 7043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404518893379205e-06,
+      "loss": 0.8653,
+      "step": 7044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.401868112351482e-06,
+      "loss": 0.9457,
+      "step": 7045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399217527086192e-06,
+      "loss": 0.8488,
+      "step": 7046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.39656713778304e-06,
+      "loss": 0.872,
+      "step": 7047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.393916944641734e-06,
+      "loss": 0.8491,
+      "step": 7048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3912669478619555e-06,
+      "loss": 0.9295,
+      "step": 7049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.388617147643371e-06,
+      "loss": 0.5206,
+      "step": 7050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.385967544185642e-06,
+      "loss": 0.5117,
+      "step": 7051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383318137688403e-06,
+      "loss": 0.8908,
+      "step": 7052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3806689283512886e-06,
+      "loss": 0.9103,
+      "step": 7053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3780199163739e-06,
+      "loss": 0.8805,
+      "step": 7054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375371101955842e-06,
+      "loss": 0.8449,
+      "step": 7055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372722485296688e-06,
+      "loss": 0.8975,
+      "step": 7056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.370074066596011e-06,
+      "loss": 0.8553,
+      "step": 7057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.36742584605336e-06,
+      "loss": 0.9039,
+      "step": 7058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364777823868266e-06,
+      "loss": 0.9279,
+      "step": 7059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3621300002402615e-06,
+      "loss": 0.8919,
+      "step": 7060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.359482375368843e-06,
+      "loss": 0.826,
+      "step": 7061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356834949453511e-06,
+      "loss": 0.8713,
+      "step": 7062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354187722693731e-06,
+      "loss": 0.9099,
+      "step": 7063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351540695288977e-06,
+      "loss": 0.8707,
+      "step": 7064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.348893867438684e-06,
+      "loss": 0.8907,
+      "step": 7065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346247239342294e-06,
+      "loss": 0.9146,
+      "step": 7066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3436008111992145e-06,
+      "loss": 0.8642,
+      "step": 7067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.340954583208854e-06,
+      "loss": 0.872,
+      "step": 7068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3383085555705965e-06,
+      "loss": 0.8904,
+      "step": 7069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.335662728483808e-06,
+      "loss": 0.9318,
+      "step": 7070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.333017102147855e-06,
+      "loss": 0.9108,
+      "step": 7071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3303716767620665e-06,
+      "loss": 0.8792,
+      "step": 7072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.327726452525779e-06,
+      "loss": 0.877,
+      "step": 7073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.325081429638294e-06,
+      "loss": 0.8505,
+      "step": 7074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.322436608298917e-06,
+      "loss": 0.8521,
+      "step": 7075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.31979198870692e-06,
+      "loss": 0.6026,
+      "step": 7076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.317147571061575e-06,
+      "loss": 0.891,
+      "step": 7077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.314503355562124e-06,
+      "loss": 0.8921,
+      "step": 7078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.311859342407809e-06,
+      "loss": 0.8633,
+      "step": 7079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.309215531797844e-06,
+      "loss": 0.9208,
+      "step": 7080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.306571923931436e-06,
+      "loss": 0.9011,
+      "step": 7081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.303928519007776e-06,
+      "loss": 0.8811,
+      "step": 7082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3012853172260315e-06,
+      "loss": 0.8683,
+      "step": 7083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2986423187853675e-06,
+      "loss": 0.9165,
+      "step": 7084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.295999523884921e-06,
+      "loss": 0.8824,
+      "step": 7085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.293356932723827e-06,
+      "loss": 0.5882,
+      "step": 7086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.290714545501189e-06,
+      "loss": 0.9405,
+      "step": 7087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288072362416112e-06,
+      "loss": 0.886,
+      "step": 7088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285430383667671e-06,
+      "loss": 0.8937,
+      "step": 7089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2827886094549385e-06,
+      "loss": 0.9084,
+      "step": 7090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280147039976963e-06,
+      "loss": 0.8868,
+      "step": 7091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2775056754327726e-06,
+      "loss": 0.868,
+      "step": 7092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2748645160213995e-06,
+      "loss": 0.8629,
+      "step": 7093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.272223561941837e-06,
+      "loss": 0.8437,
+      "step": 7094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2695828133930835e-06,
+      "loss": 0.8761,
+      "step": 7095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266942270574103e-06,
+      "loss": 0.9347,
+      "step": 7096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.264301933683864e-06,
+      "loss": 0.8714,
+      "step": 7097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.261661802921298e-06,
+      "loss": 0.9489,
+      "step": 7098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.259021878485344e-06,
+      "loss": 0.8831,
+      "step": 7099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.256382160574902e-06,
+      "loss": 0.8802,
+      "step": 7100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.253742649388874e-06,
+      "loss": 0.8508,
+      "step": 7101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251103345126138e-06,
+      "loss": 0.8715,
+      "step": 7102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.248464247985558e-06,
+      "loss": 0.893,
+      "step": 7103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24582535816599e-06,
+      "loss": 0.854,
+      "step": 7104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.243186675866256e-06,
+      "loss": 0.8583,
+      "step": 7105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240548201285186e-06,
+      "loss": 0.8354,
+      "step": 7106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23790993462157e-06,
+      "loss": 0.859,
+      "step": 7107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.235271876074206e-06,
+      "loss": 0.9487,
+      "step": 7108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232634025841853e-06,
+      "loss": 0.5245,
+      "step": 7109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.229996384123279e-06,
+      "loss": 0.9009,
+      "step": 7110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227358951117216e-06,
+      "loss": 0.9138,
+      "step": 7111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.224721727022384e-06,
+      "loss": 0.913,
+      "step": 7112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222084712037499e-06,
+      "loss": 0.8863,
+      "step": 7113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219447906361246e-06,
+      "loss": 0.9089,
+      "step": 7114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2168113101923085e-06,
+      "loss": 0.8975,
+      "step": 7115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21417492372934e-06,
+      "loss": 0.8228,
+      "step": 7116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.211538747170992e-06,
+      "loss": 0.9177,
+      "step": 7117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.208902780715888e-06,
+      "loss": 0.848,
+      "step": 7118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206267024562646e-06,
+      "loss": 0.515,
+      "step": 7119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.203631478909857e-06,
+      "loss": 0.8523,
+      "step": 7120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200996143956111e-06,
+      "loss": 0.8948,
+      "step": 7121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198361019899966e-06,
+      "loss": 0.907,
+      "step": 7122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1957261069399745e-06,
+      "loss": 0.8886,
+      "step": 7123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.193091405274671e-06,
+      "loss": 0.8841,
+      "step": 7124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1904569151025694e-06,
+      "loss": 0.5887,
+      "step": 7125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.187822636622179e-06,
+      "loss": 0.8436,
+      "step": 7126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.185188570031979e-06,
+      "loss": 0.8749,
+      "step": 7127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.182554715530446e-06,
+      "loss": 0.8883,
+      "step": 7128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.179921073316025e-06,
+      "loss": 0.8386,
+      "step": 7129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177287643587164e-06,
+      "loss": 0.8623,
+      "step": 7130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174654426542275e-06,
+      "loss": 0.8946,
+      "step": 7131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.172021422379776e-06,
+      "loss": 0.8349,
+      "step": 7132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1693886312980475e-06,
+      "loss": 0.8957,
+      "step": 7133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.166756053495463e-06,
+      "loss": 0.8931,
+      "step": 7134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.164123689170388e-06,
+      "loss": 0.8848,
+      "step": 7135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.161491538521154e-06,
+      "loss": 0.9178,
+      "step": 7136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158859601746098e-06,
+      "loss": 0.8548,
+      "step": 7137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.156227879043518e-06,
+      "loss": 0.8942,
+      "step": 7138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.153596370611719e-06,
+      "loss": 0.864,
+      "step": 7139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.150965076648968e-06,
+      "loss": 0.9146,
+      "step": 7140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.148333997353534e-06,
+      "loss": 0.8789,
+      "step": 7141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.145703132923657e-06,
+      "loss": 0.8488,
+      "step": 7142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.143072483557567e-06,
+      "loss": 0.8622,
+      "step": 7143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1404420494534775e-06,
+      "loss": 0.8568,
+      "step": 7144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1378118308095835e-06,
+      "loss": 0.9158,
+      "step": 7145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.135181827824065e-06,
+      "loss": 0.868,
+      "step": 7146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.132552040695087e-06,
+      "loss": 0.8725,
+      "step": 7147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.129922469620798e-06,
+      "loss": 0.8739,
+      "step": 7148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.127293114799325e-06,
+      "loss": 0.9147,
+      "step": 7149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.12466397642879e-06,
+      "loss": 0.8487,
+      "step": 7150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122035054707283e-06,
+      "loss": 0.8687,
+      "step": 7151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119406349832896e-06,
+      "loss": 0.8798,
+      "step": 7152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1167778620036886e-06,
+      "loss": 0.8923,
+      "step": 7153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.114149591417709e-06,
+      "loss": 0.8732,
+      "step": 7154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111521538272997e-06,
+      "loss": 0.8267,
+      "step": 7155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.108893702767561e-06,
+      "loss": 0.5694,
+      "step": 7156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.106266085099412e-06,
+      "loss": 0.9257,
+      "step": 7157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.103638685466522e-06,
+      "loss": 0.9182,
+      "step": 7158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10101150406687e-06,
+      "loss": 0.854,
+      "step": 7159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0983845410984e-06,
+      "loss": 0.931,
+      "step": 7160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.095757796759052e-06,
+      "loss": 0.8915,
+      "step": 7161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093131271246737e-06,
+      "loss": 0.8726,
+      "step": 7162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.090504964759366e-06,
+      "loss": 0.8771,
+      "step": 7163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087878877494817e-06,
+      "loss": 0.5276,
+      "step": 7164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.085253009650962e-06,
+      "loss": 0.8645,
+      "step": 7165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082627361425652e-06,
+      "loss": 0.8744,
+      "step": 7166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.080001933016724e-06,
+      "loss": 0.9499,
+      "step": 7167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.077376724621996e-06,
+      "loss": 0.8516,
+      "step": 7168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0747517364392694e-06,
+      "loss": 0.9418,
+      "step": 7169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.072126968666335e-06,
+      "loss": 0.8854,
+      "step": 7170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.069502421500956e-06,
+      "loss": 0.908,
+      "step": 7171
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066878095140892e-06,
+      "loss": 0.8729,
+      "step": 7172
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.06425398978387e-06,
+      "loss": 0.8765,
+      "step": 7173
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061630105627619e-06,
+      "loss": 0.8732,
+      "step": 7174
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.05900644286984e-06,
+      "loss": 0.8729,
+      "step": 7175
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0563830017082106e-06,
+      "loss": 0.8845,
+      "step": 7176
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.05375978234041e-06,
+      "loss": 0.8731,
+      "step": 7177
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.051136784964083e-06,
+      "loss": 0.8371,
+      "step": 7178
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.048514009776875e-06,
+      "loss": 0.8542,
+      "step": 7179
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0458914569763945e-06,
+      "loss": 0.9074,
+      "step": 7180
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.043269126760255e-06,
+      "loss": 0.9147,
+      "step": 7181
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040647019326033e-06,
+      "loss": 0.8795,
+      "step": 7182
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038025134871302e-06,
+      "loss": 0.8909,
+      "step": 7183
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035403473593614e-06,
+      "loss": 0.9121,
+      "step": 7184
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.032782035690502e-06,
+      "loss": 0.8812,
+      "step": 7185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.030160821359486e-06,
+      "loss": 0.9389,
+      "step": 7186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027539830798069e-06,
+      "loss": 0.8836,
+      "step": 7187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0249190642037334e-06,
+      "loss": 0.8991,
+      "step": 7188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022298521773949e-06,
+      "loss": 0.8514,
+      "step": 7189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019678203706164e-06,
+      "loss": 0.8668,
+      "step": 7190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017058110197814e-06,
+      "loss": 0.9076,
+      "step": 7191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.01443824144632e-06,
+      "loss": 0.8893,
+      "step": 7192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011818597649074e-06,
+      "loss": 0.8684,
+      "step": 7193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00919917900347e-06,
+      "loss": 0.8852,
+      "step": 7194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.006579985706867e-06,
+      "loss": 0.8719,
+      "step": 7195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.003961017956611e-06,
+      "loss": 0.8364,
+      "step": 7196
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001342275950045e-06,
+      "loss": 0.8891,
+      "step": 7197
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9987237598844726e-06,
+      "loss": 0.8789,
+      "step": 7198
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996105469957204e-06,
+      "loss": 0.9021,
+      "step": 7199
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.993487406365509e-06,
+      "loss": 0.8998,
+      "step": 7200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.990869569306662e-06,
+      "loss": 0.8755,
+      "step": 7201
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9882519589779005e-06,
+      "loss": 0.555,
+      "step": 7202
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.985634575576463e-06,
+      "loss": 0.8655,
+      "step": 7203
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983017419299557e-06,
+      "loss": 0.848,
+      "step": 7204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.980400490344383e-06,
+      "loss": 0.9004,
+      "step": 7205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9777837889081154e-06,
+      "loss": 0.906,
+      "step": 7206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975167315187918e-06,
+      "loss": 0.9113,
+      "step": 7207
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.972551069380935e-06,
+      "loss": 0.8713,
+      "step": 7208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969935051684293e-06,
+      "loss": 0.8883,
+      "step": 7209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967319262295101e-06,
+      "loss": 0.8895,
+      "step": 7210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.964703701410455e-06,
+      "loss": 0.9112,
+      "step": 7211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.962088369227429e-06,
+      "loss": 0.8717,
+      "step": 7212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959473265943079e-06,
+      "loss": 0.891,
+      "step": 7213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.956858391754453e-06,
+      "loss": 0.8525,
+      "step": 7214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.954243746858566e-06,
+      "loss": 0.8815,
+      "step": 7215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951629331452433e-06,
+      "loss": 0.9004,
+      "step": 7216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.94901514573304e-06,
+      "loss": 0.8452,
+      "step": 7217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9464011898973536e-06,
+      "loss": 0.8691,
+      "step": 7218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.943787464142339e-06,
+      "loss": 0.878,
+      "step": 7219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.941173968664923e-06,
+      "loss": 0.8805,
+      "step": 7220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.938560703662036e-06,
+      "loss": 0.89,
+      "step": 7221
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93594766933057e-06,
+      "loss": 0.9221,
+      "step": 7222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93333486586742e-06,
+      "loss": 0.9274,
+      "step": 7223
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.930722293469445e-06,
+      "loss": 0.9239,
+      "step": 7224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.928109952333506e-06,
+      "loss": 0.9206,
+      "step": 7225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9254978426564256e-06,
+      "loss": 0.8529,
+      "step": 7226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.922885964635025e-06,
+      "loss": 0.8921,
+      "step": 7227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.920274318466103e-06,
+      "loss": 0.893,
+      "step": 7228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9176629043464364e-06,
+      "loss": 0.9323,
+      "step": 7229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91505172247279e-06,
+      "loss": 0.858,
+      "step": 7230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912440773041911e-06,
+      "loss": 0.8771,
+      "step": 7231
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.9247,
+      "step": 7232
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9072195722953485e-06,
+      "loss": 0.9,
+      "step": 7233
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.904609321373068e-06,
+      "loss": 0.8765,
+      "step": 7234
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.901999303680359e-06,
+      "loss": 0.8458,
+      "step": 7235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899389519413886e-06,
+      "loss": 0.8972,
+      "step": 7236
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.896779968770287e-06,
+      "loss": 0.858,
+      "step": 7237
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8941706519461785e-06,
+      "loss": 0.9005,
+      "step": 7238
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.891561569138174e-06,
+      "loss": 0.9164,
+      "step": 7239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.888952720542853e-06,
+      "loss": 0.8548,
+      "step": 7240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886344106356795e-06,
+      "loss": 0.8766,
+      "step": 7241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.883735726776541e-06,
+      "loss": 0.8881,
+      "step": 7242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.881127581998639e-06,
+      "loss": 0.865,
+      "step": 7243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878519672219592e-06,
+      "loss": 0.8921,
+      "step": 7244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8759119976359116e-06,
+      "loss": 0.9052,
+      "step": 7245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.873304558444067e-06,
+      "loss": 0.8687,
+      "step": 7246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.870697354840534e-06,
+      "loss": 0.8953,
+      "step": 7247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.86809038702175e-06,
+      "loss": 0.8942,
+      "step": 7248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.865483655184146e-06,
+      "loss": 0.8622,
+      "step": 7249
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.862877159524133e-06,
+      "loss": 0.9296,
+      "step": 7250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.860270900238102e-06,
+      "loss": 0.9038,
+      "step": 7251
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8576648775224296e-06,
+      "loss": 0.8752,
+      "step": 7252
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855059091573472e-06,
+      "loss": 0.8801,
+      "step": 7253
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.852453542587569e-06,
+      "loss": 0.8919,
+      "step": 7254
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84984823076104e-06,
+      "loss": 0.8728,
+      "step": 7255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.847243156290191e-06,
+      "loss": 0.8905,
+      "step": 7256
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.844638319371303e-06,
+      "loss": 0.8891,
+      "step": 7257
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.842033720200651e-06,
+      "loss": 0.8738,
+      "step": 7258
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.839429358974481e-06,
+      "loss": 0.8323,
+      "step": 7259
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.83682523588902e-06,
+      "loss": 0.8976,
+      "step": 7260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.834221351140491e-06,
+      "loss": 0.858,
+      "step": 7261
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.83161770492508e-06,
+      "loss": 0.9035,
+      "step": 7262
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.829014297438974e-06,
+      "loss": 0.8841,
+      "step": 7263
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.826411128878325e-06,
+      "loss": 0.8794,
+      "step": 7264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8238081994392836e-06,
+      "loss": 0.9,
+      "step": 7265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8212055093179656e-06,
+      "loss": 0.895,
+      "step": 7266
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.818603058710482e-06,
+      "loss": 0.8854,
+      "step": 7267
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.81600084781292e-06,
+      "loss": 0.8838,
+      "step": 7268
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8133988768213446e-06,
+      "loss": 0.8351,
+      "step": 7269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8107971459318135e-06,
+      "loss": 0.9263,
+      "step": 7270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808195655340357e-06,
+      "loss": 0.8871,
+      "step": 7271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805594405242992e-06,
+      "loss": 0.8544,
+      "step": 7272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8029933958357155e-06,
+      "loss": 0.8409,
+      "step": 7273
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.800392627314506e-06,
+      "loss": 0.8811,
+      "step": 7274
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797792099875325e-06,
+      "loss": 0.8753,
+      "step": 7275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.795191813714117e-06,
+      "loss": 0.9071,
+      "step": 7276
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792591769026804e-06,
+      "loss": 0.8698,
+      "step": 7277
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.789991966009295e-06,
+      "loss": 0.9239,
+      "step": 7278
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78739240485748e-06,
+      "loss": 0.937,
+      "step": 7279
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7847930857672205e-06,
+      "loss": 0.8539,
+      "step": 7280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782194008934379e-06,
+      "loss": 0.8695,
+      "step": 7281
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.779595174554782e-06,
+      "loss": 0.9232,
+      "step": 7282
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7769965828242505e-06,
+      "loss": 0.8744,
+      "step": 7283
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774398233938572e-06,
+      "loss": 0.9232,
+      "step": 7284
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7718001280935384e-06,
+      "loss": 0.8595,
+      "step": 7285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.769202265484899e-06,
+      "loss": 0.8901,
+      "step": 7286
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766604646308404e-06,
+      "loss": 0.9117,
+      "step": 7287
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764007270759769e-06,
+      "loss": 0.7926,
+      "step": 7288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7614101390347095e-06,
+      "loss": 0.8461,
+      "step": 7289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758813251328907e-06,
+      "loss": 0.8823,
+      "step": 7290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.756216607838024e-06,
+      "loss": 0.8711,
+      "step": 7291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753620208757721e-06,
+      "loss": 0.8331,
+      "step": 7292
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.751024054283624e-06,
+      "loss": 0.8653,
+      "step": 7293
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7484281446113495e-06,
+      "loss": 0.9016,
+      "step": 7294
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.745832479936492e-06,
+      "loss": 0.8486,
+      "step": 7295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743237060454626e-06,
+      "loss": 0.8433,
+      "step": 7296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.740641886361312e-06,
+      "loss": 0.8828,
+      "step": 7297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.738046957852089e-06,
+      "loss": 0.8841,
+      "step": 7298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.735452275122476e-06,
+      "loss": 0.9378,
+      "step": 7299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7328578383679775e-06,
+      "loss": 0.8624,
+      "step": 7300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.730263647784079e-06,
+      "loss": 0.8526,
+      "step": 7301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7276697035662406e-06,
+      "loss": 0.9088,
+      "step": 7302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.725076005909916e-06,
+      "loss": 0.882,
+      "step": 7303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.722482555010528e-06,
+      "loss": 0.5228,
+      "step": 7304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.71988935106349e-06,
+      "loss": 0.9379,
+      "step": 7305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.717296394264188e-06,
+      "loss": 0.8532,
+      "step": 7306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714703684808004e-06,
+      "loss": 0.8958,
+      "step": 7307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.71211122289028e-06,
+      "loss": 0.8714,
+      "step": 7308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.709519008706362e-06,
+      "loss": 0.8586,
+      "step": 7309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.706927042451561e-06,
+      "loss": 0.9089,
+      "step": 7310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704335324321171e-06,
+      "loss": 0.885,
+      "step": 7311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.701743854510482e-06,
+      "loss": 0.9052,
+      "step": 7312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699152633214743e-06,
+      "loss": 0.9005,
+      "step": 7313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.696561660629203e-06,
+      "loss": 0.8415,
+      "step": 7314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693970936949084e-06,
+      "loss": 0.9099,
+      "step": 7315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.691380462369588e-06,
+      "loss": 0.8072,
+      "step": 7316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6887902370859004e-06,
+      "loss": 0.8716,
+      "step": 7317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.68620026129319e-06,
+      "loss": 0.9584,
+      "step": 7318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.683610535186604e-06,
+      "loss": 0.8686,
+      "step": 7319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.681021058961272e-06,
+      "loss": 0.8799,
+      "step": 7320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.678431832812305e-06,
+      "loss": 0.8557,
+      "step": 7321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.675842856934789e-06,
+      "loss": 0.832,
+      "step": 7322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.673254131523805e-06,
+      "loss": 0.8651,
+      "step": 7323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670665656774398e-06,
+      "loss": 0.8481,
+      "step": 7324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66807743288161e-06,
+      "loss": 0.9101,
+      "step": 7325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66548946004045e-06,
+      "loss": 0.5217,
+      "step": 7326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.662901738445923e-06,
+      "loss": 0.8973,
+      "step": 7327
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660314268293e-06,
+      "loss": 0.8993,
+      "step": 7328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.657727049776645e-06,
+      "loss": 0.8805,
+      "step": 7329
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.655140083091794e-06,
+      "loss": 0.5562,
+      "step": 7330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6525533684333724e-06,
+      "loss": 0.8807,
+      "step": 7331
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.649966905996281e-06,
+      "loss": 0.8581,
+      "step": 7332
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647380695975398e-06,
+      "loss": 0.8188,
+      "step": 7333
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644794738565597e-06,
+      "loss": 0.9042,
+      "step": 7334
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.642209033961712e-06,
+      "loss": 0.8968,
+      "step": 7335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.639623582358577e-06,
+      "loss": 0.9197,
+      "step": 7336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637038383950998e-06,
+      "loss": 0.9257,
+      "step": 7337
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.634453438933761e-06,
+      "loss": 0.9081,
+      "step": 7338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631868747501636e-06,
+      "loss": 0.8459,
+      "step": 7339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629284309849373e-06,
+      "loss": 0.9296,
+      "step": 7340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6267001261717015e-06,
+      "loss": 0.8806,
+      "step": 7341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.624116196663332e-06,
+      "loss": 0.8782,
+      "step": 7342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.621532521518962e-06,
+      "loss": 0.8685,
+      "step": 7343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618949100933256e-06,
+      "loss": 0.9013,
+      "step": 7344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.616365935100877e-06,
+      "loss": 0.9332,
+      "step": 7345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.613783024216451e-06,
+      "loss": 0.9077,
+      "step": 7346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611200368474604e-06,
+      "loss": 0.9239,
+      "step": 7347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.608617968069921e-06,
+      "loss": 0.8971,
+      "step": 7348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60603582319699e-06,
+      "loss": 0.8398,
+      "step": 7349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.603453934050359e-06,
+      "loss": 0.9094,
+      "step": 7350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.600872300824574e-06,
+      "loss": 0.8565,
+      "step": 7351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.598290923714152e-06,
+      "loss": 0.8973,
+      "step": 7352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.595709802913588e-06,
+      "loss": 0.8589,
+      "step": 7353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.593128938617373e-06,
+      "loss": 0.8904,
+      "step": 7354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.590548331019957e-06,
+      "loss": 0.8371,
+      "step": 7355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5879679803157906e-06,
+      "loss": 0.8477,
+      "step": 7356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.58538788669929e-06,
+      "loss": 0.9138,
+      "step": 7357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582808050364864e-06,
+      "loss": 0.8983,
+      "step": 7358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.580228471506894e-06,
+      "loss": 0.8881,
+      "step": 7359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.577649150319745e-06,
+      "loss": 0.8823,
+      "step": 7360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.575070086997762e-06,
+      "loss": 0.5902,
+      "step": 7361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5724912817352705e-06,
+      "loss": 0.8633,
+      "step": 7362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5699127347265775e-06,
+      "loss": 0.8939,
+      "step": 7363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.567334446165967e-06,
+      "loss": 0.9031,
+      "step": 7364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564756416247712e-06,
+      "loss": 0.9208,
+      "step": 7365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5621786451660515e-06,
+      "loss": 0.9367,
+      "step": 7366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559601133115223e-06,
+      "loss": 0.8479,
+      "step": 7367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557023880289428e-06,
+      "loss": 0.8257,
+      "step": 7368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554446886882864e-06,
+      "loss": 0.875,
+      "step": 7369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.55187015308969e-06,
+      "loss": 0.9102,
+      "step": 7370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549293679104065e-06,
+      "loss": 0.8562,
+      "step": 7371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.546717465120114e-06,
+      "loss": 0.8819,
+      "step": 7372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544141511331954e-06,
+      "loss": 0.8726,
+      "step": 7373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541565817933672e-06,
+      "loss": 0.897,
+      "step": 7374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.538990385119336e-06,
+      "loss": 0.8825,
+      "step": 7375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.536415213083007e-06,
+      "loss": 0.8991,
+      "step": 7376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.533840302018709e-06,
+      "loss": 0.871,
+      "step": 7377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.531265652120463e-06,
+      "loss": 0.8714,
+      "step": 7378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.528691263582254e-06,
+      "loss": 0.9247,
+      "step": 7379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5261171365980616e-06,
+      "loss": 0.8827,
+      "step": 7380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.523543271361837e-06,
+      "loss": 0.8976,
+      "step": 7381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.520969668067514e-06,
+      "loss": 0.8815,
+      "step": 7382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.518396326909007e-06,
+      "loss": 0.9086,
+      "step": 7383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.515823248080211e-06,
+      "loss": 0.8404,
+      "step": 7384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.513250431775003e-06,
+      "loss": 0.8879,
+      "step": 7385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.510677878187233e-06,
+      "loss": 0.8927,
+      "step": 7386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.50810558751074e-06,
+      "loss": 0.8514,
+      "step": 7387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.505533559939335e-06,
+      "loss": 0.9118,
+      "step": 7388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5029617956668214e-06,
+      "loss": 0.8744,
+      "step": 7389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500390294886965e-06,
+      "loss": 0.9527,
+      "step": 7390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497819057793531e-06,
+      "loss": 0.8581,
+      "step": 7391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.495248084580248e-06,
+      "loss": 0.8733,
+      "step": 7392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.492677375440839e-06,
+      "loss": 0.8733,
+      "step": 7393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4901069305689955e-06,
+      "loss": 0.8912,
+      "step": 7394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487536750158391e-06,
+      "loss": 0.8761,
+      "step": 7395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484966834402689e-06,
+      "loss": 0.8801,
+      "step": 7396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.482397183495519e-06,
+      "loss": 0.88,
+      "step": 7397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479827797630506e-06,
+      "loss": 0.8717,
+      "step": 7398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.477258677001237e-06,
+      "loss": 0.9008,
+      "step": 7399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474689821801295e-06,
+      "loss": 0.8748,
+      "step": 7400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.472121232224232e-06,
+      "loss": 0.9014,
+      "step": 7401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4695529084635885e-06,
+      "loss": 0.8567,
+      "step": 7402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.466984850712881e-06,
+      "loss": 0.4942,
+      "step": 7403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.464417059165602e-06,
+      "loss": 0.8671,
+      "step": 7404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.461849534015233e-06,
+      "loss": 0.892,
+      "step": 7405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.459282275455223e-06,
+      "loss": 0.8922,
+      "step": 7406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.456715283679018e-06,
+      "loss": 0.9001,
+      "step": 7407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.454148558880023e-06,
+      "loss": 0.8807,
+      "step": 7408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.451582101251645e-06,
+      "loss": 0.8929,
+      "step": 7409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44901591098725e-06,
+      "loss": 0.8479,
+      "step": 7410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.446449988280202e-06,
+      "loss": 0.8805,
+      "step": 7411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44388433332383e-06,
+      "loss": 0.869,
+      "step": 7412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.441318946311454e-06,
+      "loss": 0.8797,
+      "step": 7413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4387538274363655e-06,
+      "loss": 0.8573,
+      "step": 7414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.436188976891846e-06,
+      "loss": 0.8765,
+      "step": 7415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4336243948711445e-06,
+      "loss": 0.5455,
+      "step": 7416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.431060081567493e-06,
+      "loss": 0.8492,
+      "step": 7417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428496037174112e-06,
+      "loss": 0.7474,
+      "step": 7418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.425932261884191e-06,
+      "loss": 0.9064,
+      "step": 7419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.423368755890909e-06,
+      "loss": 0.9023,
+      "step": 7420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.420805519387412e-06,
+      "loss": 0.883,
+      "step": 7421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418242552566841e-06,
+      "loss": 0.858,
+      "step": 7422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.415679855622302e-06,
+      "loss": 0.8406,
+      "step": 7423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.413117428746892e-06,
+      "loss": 0.8914,
+      "step": 7424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.410555272133683e-06,
+      "loss": 0.8726,
+      "step": 7425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.407993385975724e-06,
+      "loss": 0.9176,
+      "step": 7426
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405431770466051e-06,
+      "loss": 0.8993,
+      "step": 7427
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402870425797668e-06,
+      "loss": 0.822,
+      "step": 7428
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400309352163574e-06,
+      "loss": 0.8692,
+      "step": 7429
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39774854975673e-06,
+      "loss": 0.8931,
+      "step": 7430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395188018770096e-06,
+      "loss": 0.8483,
+      "step": 7431
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39262775939659e-06,
+      "loss": 0.8742,
+      "step": 7432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.390067771829132e-06,
+      "loss": 0.89,
+      "step": 7433
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3875080562606005e-06,
+      "loss": 0.8803,
+      "step": 7434
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.384948612883872e-06,
+      "loss": 0.8473,
+      "step": 7435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3823894418917895e-06,
+      "loss": 0.9511,
+      "step": 7436
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.379830543477177e-06,
+      "loss": 0.8752,
+      "step": 7437
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.377271917832848e-06,
+      "loss": 0.8595,
+      "step": 7438
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.374713565151579e-06,
+      "loss": 0.9049,
+      "step": 7439
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3721554856261445e-06,
+      "loss": 0.9014,
+      "step": 7440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.369597679449282e-06,
+      "loss": 0.8688,
+      "step": 7441
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.367040146813721e-06,
+      "loss": 0.8959,
+      "step": 7442
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.364482887912159e-06,
+      "loss": 0.8078,
+      "step": 7443
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3619259029372845e-06,
+      "loss": 0.8894,
+      "step": 7444
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.359369192081756e-06,
+      "loss": 0.8859,
+      "step": 7445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.356812755538215e-06,
+      "loss": 0.8905,
+      "step": 7446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3542565934992885e-06,
+      "loss": 0.8624,
+      "step": 7447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.351700706157565e-06,
+      "loss": 0.9101,
+      "step": 7448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.349145093705635e-06,
+      "loss": 0.855,
+      "step": 7449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34658975633605e-06,
+      "loss": 0.9385,
+      "step": 7450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.344034694241353e-06,
+      "loss": 0.8879,
+      "step": 7451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.341479907614055e-06,
+      "loss": 0.8765,
+      "step": 7452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338925396646661e-06,
+      "loss": 0.8672,
+      "step": 7453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3363711615316384e-06,
+      "loss": 0.8484,
+      "step": 7454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333817202461447e-06,
+      "loss": 0.894,
+      "step": 7455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.331263519628519e-06,
+      "loss": 0.91,
+      "step": 7456
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.328710113225271e-06,
+      "loss": 0.9046,
+      "step": 7457
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.326156983444092e-06,
+      "loss": 0.8878,
+      "step": 7458
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3236041304773525e-06,
+      "loss": 0.8988,
+      "step": 7459
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.321051554517406e-06,
+      "loss": 0.8972,
+      "step": 7460
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.318499255756581e-06,
+      "loss": 0.8342,
+      "step": 7461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31594723438719e-06,
+      "loss": 0.86,
+      "step": 7462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.313395490601513e-06,
+      "loss": 0.884,
+      "step": 7463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310844024591826e-06,
+      "loss": 0.9052,
+      "step": 7464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.308292836550371e-06,
+      "loss": 0.8785,
+      "step": 7465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.305741926669376e-06,
+      "loss": 0.9182,
+      "step": 7466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303191295141042e-06,
+      "loss": 0.8817,
+      "step": 7467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300640942157551e-06,
+      "loss": 0.8473,
+      "step": 7468
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.298090867911073e-06,
+      "loss": 0.8501,
+      "step": 7469
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2955410725937405e-06,
+      "loss": 0.9165,
+      "step": 7470
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.292991556397683e-06,
+      "loss": 0.87,
+      "step": 7471
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290442319514989e-06,
+      "loss": 0.8519,
+      "step": 7472
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287893362137749e-06,
+      "loss": 0.8436,
+      "step": 7473
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285344684458008e-06,
+      "loss": 0.8695,
+      "step": 7474
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282796286667814e-06,
+      "loss": 0.9539,
+      "step": 7475
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.280248168959171e-06,
+      "loss": 0.8551,
+      "step": 7476
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277700331524083e-06,
+      "loss": 0.8938,
+      "step": 7477
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275152774554518e-06,
+      "loss": 0.8744,
+      "step": 7478
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2726054982424235e-06,
+      "loss": 0.8715,
+      "step": 7479
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.270058502779739e-06,
+      "loss": 0.8656,
+      "step": 7480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267511788358365e-06,
+      "loss": 0.8773,
+      "step": 7481
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2649653551701985e-06,
+      "loss": 0.906,
+      "step": 7482
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2624192034070974e-06,
+      "loss": 0.8465,
+      "step": 7483
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259873333260917e-06,
+      "loss": 0.9294,
+      "step": 7484
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2573277449234735e-06,
+      "loss": 0.556,
+      "step": 7485
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254782438586579e-06,
+      "loss": 0.863,
+      "step": 7486
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.252237414442006e-06,
+      "loss": 0.8455,
+      "step": 7487
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2496926726815246e-06,
+      "loss": 0.8861,
+      "step": 7488
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247148213496867e-06,
+      "loss": 0.9104,
+      "step": 7489
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.244604037079754e-06,
+      "loss": 0.8244,
+      "step": 7490
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2420601436218855e-06,
+      "loss": 0.8736,
+      "step": 7491
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.239516533314931e-06,
+      "loss": 0.8809,
+      "step": 7492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.236973206350554e-06,
+      "loss": 0.8788,
+      "step": 7493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.234430162920378e-06,
+      "loss": 0.8979,
+      "step": 7494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.231887403216024e-06,
+      "loss": 0.8057,
+      "step": 7495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.22934492742907e-06,
+      "loss": 0.9001,
+      "step": 7496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2268027357510985e-06,
+      "loss": 0.8634,
+      "step": 7497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.224260828373647e-06,
+      "loss": 0.8767,
+      "step": 7498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.221719205488248e-06,
+      "loss": 0.8784,
+      "step": 7499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.219177867286403e-06,
+      "loss": 0.8849,
+      "step": 7500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.216636813959594e-06,
+      "loss": 0.8854,
+      "step": 7501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.214096045699285e-06,
+      "loss": 0.9238,
+      "step": 7502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.211555562696914e-06,
+      "loss": 0.8443,
+      "step": 7503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209015365143902e-06,
+      "loss": 0.908,
+      "step": 7504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206475453231644e-06,
+      "loss": 0.8548,
+      "step": 7505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2039358271515194e-06,
+      "loss": 0.8664,
+      "step": 7506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.201396487094877e-06,
+      "loss": 0.906,
+      "step": 7507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198857433253056e-06,
+      "loss": 0.8978,
+      "step": 7508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.196318665817363e-06,
+      "loss": 0.9064,
+      "step": 7509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.193780184979087e-06,
+      "loss": 0.897,
+      "step": 7510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191241990929498e-06,
+      "loss": 0.8957,
+      "step": 7511
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188704083859839e-06,
+      "loss": 0.8964,
+      "step": 7512
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18616646396134e-06,
+      "loss": 0.8907,
+      "step": 7513
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1836291314252e-06,
+      "loss": 0.8581,
+      "step": 7514
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.181092086442604e-06,
+      "loss": 0.8842,
+      "step": 7515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1785553292047055e-06,
+      "loss": 0.8411,
+      "step": 7516
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17601885990265e-06,
+      "loss": 0.8644,
+      "step": 7517
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173482678727545e-06,
+      "loss": 0.8832,
+      "step": 7518
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.170946785870495e-06,
+      "loss": 0.8628,
+      "step": 7519
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168411181522569e-06,
+      "loss": 0.5636,
+      "step": 7520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.165875865874811e-06,
+      "loss": 0.5411,
+      "step": 7521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163340839118261e-06,
+      "loss": 0.9129,
+      "step": 7522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.160806101443919e-06,
+      "loss": 0.8629,
+      "step": 7523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.158271653042779e-06,
+      "loss": 0.9264,
+      "step": 7524
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1557374941057945e-06,
+      "loss": 0.8786,
+      "step": 7525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153203624823918e-06,
+      "loss": 0.9146,
+      "step": 7526
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1506700453880604e-06,
+      "loss": 0.8337,
+      "step": 7527
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.14813675598913e-06,
+      "loss": 0.8739,
+      "step": 7528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.145603756817994e-06,
+      "loss": 0.8793,
+      "step": 7529
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.143071048065517e-06,
+      "loss": 0.8591,
+      "step": 7530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.140538629922522e-06,
+      "loss": 0.9329,
+      "step": 7531
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1380065025798275e-06,
+      "loss": 0.8799,
+      "step": 7532
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.135474666228218e-06,
+      "loss": 0.8948,
+      "step": 7533
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1329431210584624e-06,
+      "loss": 0.8683,
+      "step": 7534
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13041186726131e-06,
+      "loss": 0.8858,
+      "step": 7535
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127880905027476e-06,
+      "loss": 0.8785,
+      "step": 7536
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12535023454767e-06,
+      "loss": 0.5671,
+      "step": 7537
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.122819856012564e-06,
+      "loss": 0.9103,
+      "step": 7538
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.120289769612824e-06,
+      "loss": 0.9115,
+      "step": 7539
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.117759975539075e-06,
+      "loss": 0.8927,
+      "step": 7540
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115230473981939e-06,
+      "loss": 0.8751,
+      "step": 7541
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1127012651320065e-06,
+      "loss": 0.9058,
+      "step": 7542
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.110172349179839e-06,
+      "loss": 0.8843,
+      "step": 7543
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.107643726315993e-06,
+      "loss": 0.816,
+      "step": 7544
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105115396730985e-06,
+      "loss": 0.8771,
+      "step": 7545
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1025873606153285e-06,
+      "loss": 0.8459,
+      "step": 7546
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100059618159493e-06,
+      "loss": 0.5001,
+      "step": 7547
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.097532169553947e-06,
+      "loss": 0.9256,
+      "step": 7548
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.095005014989119e-06,
+      "loss": 0.813,
+      "step": 7549
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092478154655431e-06,
+      "loss": 0.8729,
+      "step": 7550
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08995158874327e-06,
+      "loss": 0.9079,
+      "step": 7551
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.087425317443007e-06,
+      "loss": 0.8456,
+      "step": 7552
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.084899340944993e-06,
+      "loss": 0.8522,
+      "step": 7553
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08237365943955e-06,
+      "loss": 0.827,
+      "step": 7554
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.079848273116983e-06,
+      "loss": 0.8965,
+      "step": 7555
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077323182167572e-06,
+      "loss": 0.864,
+      "step": 7556
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07479838678158e-06,
+      "loss": 0.9122,
+      "step": 7557
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.072273887149236e-06,
+      "loss": 0.8494,
+      "step": 7558
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069749683460765e-06,
+      "loss": 0.8341,
+      "step": 7559
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.067225775906349e-06,
+      "loss": 0.8813,
+      "step": 7560
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0647021646761664e-06,
+      "loss": 0.8932,
+      "step": 7561
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.062178849960359e-06,
+      "loss": 0.9088,
+      "step": 7562
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0596558319490515e-06,
+      "loss": 0.9055,
+      "step": 7563
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.057133110832351e-06,
+      "loss": 0.8753,
+      "step": 7564
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.054610686800333e-06,
+      "loss": 0.8595,
+      "step": 7565
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0520885600430615e-06,
+      "loss": 0.7933,
+      "step": 7566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049566730750564e-06,
+      "loss": 0.8845,
+      "step": 7567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.047045199112865e-06,
+      "loss": 0.8198,
+      "step": 7568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.044523965319942e-06,
+      "loss": 0.8301,
+      "step": 7569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042003029561776e-06,
+      "loss": 0.8761,
+      "step": 7570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039482392028302e-06,
+      "loss": 0.8567,
+      "step": 7571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.036962052909453e-06,
+      "loss": 0.8857,
+      "step": 7572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.034442012395123e-06,
+      "loss": 0.8545,
+      "step": 7573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.031922270675193e-06,
+      "loss": 0.8713,
+      "step": 7574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.029402827939519e-06,
+      "loss": 0.8477,
+      "step": 7575
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.026883684377935e-06,
+      "loss": 0.8979,
+      "step": 7576
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.02436484018025e-06,
+      "loss": 0.8759,
+      "step": 7577
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.021846295536251e-06,
+      "loss": 0.9262,
+      "step": 7578
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.019328050635711e-06,
+      "loss": 0.8067,
+      "step": 7579
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016810105668365e-06,
+      "loss": 0.8636,
+      "step": 7580
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.014292460823938e-06,
+      "loss": 0.8872,
+      "step": 7581
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.011775116292125e-06,
+      "loss": 0.8537,
+      "step": 7582
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.009258072262607e-06,
+      "loss": 0.5651,
+      "step": 7583
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.006741328925033e-06,
+      "loss": 0.8836,
+      "step": 7584
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0042248864690275e-06,
+      "loss": 0.8768,
+      "step": 7585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.001708745084209e-06,
+      "loss": 0.8679,
+      "step": 7586
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.999192904960151e-06,
+      "loss": 0.8588,
+      "step": 7587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.996677366286425e-06,
+      "loss": 0.9034,
+      "step": 7588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.994162129252561e-06,
+      "loss": 0.881,
+      "step": 7589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.991647194048087e-06,
+      "loss": 0.8819,
+      "step": 7590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.989132560862484e-06,
+      "loss": 0.9103,
+      "step": 7591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986618229885234e-06,
+      "loss": 0.9073,
+      "step": 7592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.984104201305778e-06,
+      "loss": 0.8996,
+      "step": 7593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.981590475313544e-06,
+      "loss": 0.9058,
+      "step": 7594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979077052097936e-06,
+      "loss": 0.856,
+      "step": 7595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976563931848332e-06,
+      "loss": 0.8673,
+      "step": 7596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.974051114754089e-06,
+      "loss": 0.8464,
+      "step": 7597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.971538601004542e-06,
+      "loss": 0.8225,
+      "step": 7598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.969026390789001e-06,
+      "loss": 0.9027,
+      "step": 7599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966514484296755e-06,
+      "loss": 0.8425,
+      "step": 7600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.964002881717073e-06,
+      "loss": 0.8824,
+      "step": 7601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.961491583239192e-06,
+      "loss": 0.889,
+      "step": 7602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.958980589052338e-06,
+      "loss": 0.633,
+      "step": 7603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.956469899345704e-06,
+      "loss": 0.8838,
+      "step": 7604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95395951430846e-06,
+      "loss": 0.846,
+      "step": 7605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951449434129766e-06,
+      "loss": 0.8901,
+      "step": 7606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94893965899874e-06,
+      "loss": 0.9382,
+      "step": 7607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946430189104497e-06,
+      "loss": 0.8471,
+      "step": 7608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94392102463611e-06,
+      "loss": 0.8543,
+      "step": 7609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.941412165782645e-06,
+      "loss": 0.86,
+      "step": 7610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.938903612733132e-06,
+      "loss": 0.8397,
+      "step": 7611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.93639536567659e-06,
+      "loss": 0.8895,
+      "step": 7612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.933887424802003e-06,
+      "loss": 0.9123,
+      "step": 7613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.931379790298344e-06,
+      "loss": 0.876,
+      "step": 7614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.928872462354551e-06,
+      "loss": 0.5041,
+      "step": 7615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.926365441159547e-06,
+      "loss": 0.8616,
+      "step": 7616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9238587269022296e-06,
+      "loss": 0.8696,
+      "step": 7617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.921352319771472e-06,
+      "loss": 0.8494,
+      "step": 7618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.918846219956126e-06,
+      "loss": 0.9118,
+      "step": 7619
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.91634042764502e-06,
+      "loss": 0.8607,
+      "step": 7620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.91383494302696e-06,
+      "loss": 0.8849,
+      "step": 7621
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.911329766290723e-06,
+      "loss": 0.8703,
+      "step": 7622
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.908824897625075e-06,
+      "loss": 0.8755,
+      "step": 7623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.906320337218743e-06,
+      "loss": 0.8435,
+      "step": 7624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903816085260447e-06,
+      "loss": 0.8411,
+      "step": 7625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.901312141938874e-06,
+      "loss": 0.8796,
+      "step": 7626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898808507442683e-06,
+      "loss": 0.9045,
+      "step": 7627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.896305181960524e-06,
+      "loss": 0.8518,
+      "step": 7628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89380216568101e-06,
+      "loss": 0.8376,
+      "step": 7629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.891299458792743e-06,
+      "loss": 0.8338,
+      "step": 7630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888797061484288e-06,
+      "loss": 0.8771,
+      "step": 7631
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886294973944202e-06,
+      "loss": 0.9122,
+      "step": 7632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883793196361004e-06,
+      "loss": 0.8476,
+      "step": 7633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.881291728923202e-06,
+      "loss": 0.9085,
+      "step": 7634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878790571819272e-06,
+      "loss": 0.8639,
+      "step": 7635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876289725237668e-06,
+      "loss": 0.8773,
+      "step": 7636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8737891893668255e-06,
+      "loss": 0.8812,
+      "step": 7637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.871288964395152e-06,
+      "loss": 0.8975,
+      "step": 7638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868789050511032e-06,
+      "loss": 0.8476,
+      "step": 7639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866289447902829e-06,
+      "loss": 0.8611,
+      "step": 7640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.86379015675888e-06,
+      "loss": 0.8987,
+      "step": 7641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.861291177267502e-06,
+      "loss": 0.8125,
+      "step": 7642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858792509616984e-06,
+      "loss": 0.8521,
+      "step": 7643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.856294153995595e-06,
+      "loss": 0.8776,
+      "step": 7644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853796110591583e-06,
+      "loss": 0.888,
+      "step": 7645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8512983795931665e-06,
+      "loss": 0.8779,
+      "step": 7646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848800961188539e-06,
+      "loss": 0.8896,
+      "step": 7647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.846303855565883e-06,
+      "loss": 0.8968,
+      "step": 7648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.843807062913338e-06,
+      "loss": 0.88,
+      "step": 7649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.841310583419041e-06,
+      "loss": 0.8966,
+      "step": 7650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838814417271087e-06,
+      "loss": 0.874,
+      "step": 7651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.836318564657561e-06,
+      "loss": 0.8503,
+      "step": 7652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833823025766518e-06,
+      "loss": 0.8393,
+      "step": 7653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8313278007859886e-06,
+      "loss": 0.8784,
+      "step": 7654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828832889903983e-06,
+      "loss": 0.8622,
+      "step": 7655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.826338293308486e-06,
+      "loss": 0.8581,
+      "step": 7656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8238440111874625e-06,
+      "loss": 0.867,
+      "step": 7657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82135004372884e-06,
+      "loss": 0.9125,
+      "step": 7658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818856391120544e-06,
+      "loss": 0.8955,
+      "step": 7659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816363053550453e-06,
+      "loss": 0.8201,
+      "step": 7660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813870031206448e-06,
+      "loss": 0.926,
+      "step": 7661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811377324276354e-06,
+      "loss": 0.8577,
+      "step": 7662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.808884932948008e-06,
+      "loss": 0.868,
+      "step": 7663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806392857409189e-06,
+      "loss": 0.9204,
+      "step": 7664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.803901097847684e-06,
+      "loss": 0.9197,
+      "step": 7665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801409654451224e-06,
+      "loss": 0.8939,
+      "step": 7666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798918527407549e-06,
+      "loss": 0.9166,
+      "step": 7667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.796427716904347e-06,
+      "loss": 0.8694,
+      "step": 7668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7939372231293e-06,
+      "loss": 0.9111,
+      "step": 7669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.791447046270055e-06,
+      "loss": 0.8972,
+      "step": 7670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.788957186514247e-06,
+      "loss": 0.903,
+      "step": 7671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.786467644049476e-06,
+      "loss": 0.8863,
+      "step": 7672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.783978419063323e-06,
+      "loss": 0.8425,
+      "step": 7673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.781489511743345e-06,
+      "loss": 0.5357,
+      "step": 7674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.779000922277076e-06,
+      "loss": 0.8643,
+      "step": 7675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7765126508520216e-06,
+      "loss": 0.877,
+      "step": 7676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.77402469765567e-06,
+      "loss": 0.9044,
+      "step": 7677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.771537062875481e-06,
+      "loss": 0.903,
+      "step": 7678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.769049746698889e-06,
+      "loss": 0.8349,
+      "step": 7679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.766562749313309e-06,
+      "loss": 0.8833,
+      "step": 7680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764076070906128e-06,
+      "loss": 0.8584,
+      "step": 7681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.761589711664714e-06,
+      "loss": 0.8825,
+      "step": 7682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.759103671776405e-06,
+      "loss": 0.8808,
+      "step": 7683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756617951428517e-06,
+      "loss": 0.8928,
+      "step": 7684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.754132550808345e-06,
+      "loss": 0.8931,
+      "step": 7685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.751647470103155e-06,
+      "loss": 0.9063,
+      "step": 7686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7491627095001925e-06,
+      "loss": 0.864,
+      "step": 7687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.746678269186682e-06,
+      "loss": 0.88,
+      "step": 7688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.744194149349807e-06,
+      "loss": 0.8912,
+      "step": 7689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741710350176751e-06,
+      "loss": 0.8864,
+      "step": 7690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.739226871854659e-06,
+      "loss": 0.8516,
+      "step": 7691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736743714570654e-06,
+      "loss": 0.9121,
+      "step": 7692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.734260878511835e-06,
+      "loss": 0.8724,
+      "step": 7693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.731778363865278e-06,
+      "loss": 0.8475,
+      "step": 7694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.729296170818033e-06,
+      "loss": 0.8849,
+      "step": 7695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.726814299557128e-06,
+      "loss": 0.878,
+      "step": 7696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724332750269563e-06,
+      "loss": 0.8814,
+      "step": 7697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.721851523142319e-06,
+      "loss": 0.8563,
+      "step": 7698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719370618362353e-06,
+      "loss": 0.8321,
+      "step": 7699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.716890036116582e-06,
+      "loss": 0.8845,
+      "step": 7700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714409776591929e-06,
+      "loss": 0.8804,
+      "step": 7701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711929839975257e-06,
+      "loss": 0.84,
+      "step": 7702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709450226453439e-06,
+      "loss": 0.8842,
+      "step": 7703
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.706970936213294e-06,
+      "loss": 0.877,
+      "step": 7704
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.704491969441643e-06,
+      "loss": 0.884,
+      "step": 7705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.702013326325256e-06,
+      "loss": 0.8207,
+      "step": 7706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699535007050905e-06,
+      "loss": 0.8363,
+      "step": 7707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6970570118053116e-06,
+      "loss": 0.8638,
+      "step": 7708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694579340775202e-06,
+      "loss": 0.9003,
+      "step": 7709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.69210199414725e-06,
+      "loss": 0.8535,
+      "step": 7710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68962497210812e-06,
+      "loss": 0.8253,
+      "step": 7711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.687148274844449e-06,
+      "loss": 0.9001,
+      "step": 7712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.684671902542853e-06,
+      "loss": 0.9078,
+      "step": 7713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.682195855389917e-06,
+      "loss": 0.8707,
+      "step": 7714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6797201335722064e-06,
+      "loss": 0.8547,
+      "step": 7715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.677244737276258e-06,
+      "loss": 0.8388,
+      "step": 7716
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.674769666688589e-06,
+      "loss": 0.9145,
+      "step": 7717
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672294921995687e-06,
+      "loss": 0.9115,
+      "step": 7718
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.669820503384019e-06,
+      "loss": 0.829,
+      "step": 7719
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.667346411040026e-06,
+      "loss": 0.821,
+      "step": 7720
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664872645150126e-06,
+      "loss": 0.8837,
+      "step": 7721
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6623992059007075e-06,
+      "loss": 0.8628,
+      "step": 7722
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6599260934781386e-06,
+      "loss": 0.9045,
+      "step": 7723
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.657453308068763e-06,
+      "loss": 0.894,
+      "step": 7724
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6549808498588986e-06,
+      "loss": 0.8545,
+      "step": 7725
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.652508719034837e-06,
+      "loss": 0.859,
+      "step": 7726
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650036915782849e-06,
+      "loss": 0.8358,
+      "step": 7727
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647565440289177e-06,
+      "loss": 0.8686,
+      "step": 7728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6450942927400396e-06,
+      "loss": 0.9002,
+      "step": 7729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.642623473321638e-06,
+      "loss": 0.8553,
+      "step": 7730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.640152982220126e-06,
+      "loss": 0.9259,
+      "step": 7731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.637682819621668e-06,
+      "loss": 0.8578,
+      "step": 7732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.635212985712366e-06,
+      "loss": 0.8713,
+      "step": 7733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.632743480678331e-06,
+      "loss": 0.8911,
+      "step": 7734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.630274304705625e-06,
+      "loss": 0.9396,
+      "step": 7735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.627805457980298e-06,
+      "loss": 0.8562,
+      "step": 7736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6253369406883675e-06,
+      "loss": 0.5559,
+      "step": 7737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.622868753015833e-06,
+      "loss": 0.8321,
+      "step": 7738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6204008951486636e-06,
+      "loss": 0.9234,
+      "step": 7739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617933367272808e-06,
+      "loss": 0.8984,
+      "step": 7740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.61546616957419e-06,
+      "loss": 0.8859,
+      "step": 7741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612999302238696e-06,
+      "loss": 0.905,
+      "step": 7742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610532765452215e-06,
+      "loss": 0.8769,
+      "step": 7743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.608066559400574e-06,
+      "loss": 0.8641,
+      "step": 7744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6056006842696145e-06,
+      "loss": 0.8852,
+      "step": 7745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6031351402451175e-06,
+      "loss": 0.864,
+      "step": 7746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6006699275128694e-06,
+      "loss": 0.8398,
+      "step": 7747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.598205046258603e-06,
+      "loss": 0.8978,
+      "step": 7748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5957404966680574e-06,
+      "loss": 0.9058,
+      "step": 7749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593276278926912e-06,
+      "loss": 0.8564,
+      "step": 7750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5908123932208565e-06,
+      "loss": 0.8458,
+      "step": 7751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588348839735527e-06,
+      "loss": 0.8374,
+      "step": 7752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5858856186565466e-06,
+      "loss": 0.8766,
+      "step": 7753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5834227301695166e-06,
+      "loss": 0.4971,
+      "step": 7754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.580960174460006e-06,
+      "loss": 0.8526,
+      "step": 7755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5784979517135625e-06,
+      "loss": 0.856,
+      "step": 7756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.576036062115709e-06,
+      "loss": 0.5753,
+      "step": 7757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5735745058519445e-06,
+      "loss": 0.895,
+      "step": 7758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571113283107735e-06,
+      "loss": 0.8808,
+      "step": 7759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.568652394068532e-06,
+      "loss": 0.5162,
+      "step": 7760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.566191838919757e-06,
+      "loss": 0.8174,
+      "step": 7761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.563731617846804e-06,
+      "loss": 0.853,
+      "step": 7762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.561271731035045e-06,
+      "loss": 0.9108,
+      "step": 7763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.558812178669827e-06,
+      "loss": 0.8659,
+      "step": 7764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55635296093647e-06,
+      "loss": 0.8672,
+      "step": 7765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55389407802027e-06,
+      "loss": 0.8534,
+      "step": 7766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.551435530106496e-06,
+      "loss": 0.8671,
+      "step": 7767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.548977317380395e-06,
+      "loss": 0.8869,
+      "step": 7768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.546519440027186e-06,
+      "loss": 0.8746,
+      "step": 7769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.544061898232064e-06,
+      "loss": 0.9197,
+      "step": 7770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.541604692180198e-06,
+      "loss": 0.8796,
+      "step": 7771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539147822056736e-06,
+      "loss": 0.8771,
+      "step": 7772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.536691288046785e-06,
+      "loss": 0.8638,
+      "step": 7773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.534235090335453e-06,
+      "loss": 0.9036,
+      "step": 7774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.531779229107797e-06,
+      "loss": 0.8883,
+      "step": 7775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529323704548868e-06,
+      "loss": 0.9178,
+      "step": 7776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.526868516843673e-06,
+      "loss": 0.8708,
+      "step": 7777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524413666177216e-06,
+      "loss": 0.8856,
+      "step": 7778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521959152734455e-06,
+      "loss": 0.8752,
+      "step": 7779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.519504976700335e-06,
+      "loss": 0.8815,
+      "step": 7780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.517051138259771e-06,
+      "loss": 0.8712,
+      "step": 7781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.514597637597654e-06,
+      "loss": 0.8529,
+      "step": 7782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5121444748988505e-06,
+      "loss": 0.8595,
+      "step": 7783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.50969165034819e-06,
+      "loss": 0.9172,
+      "step": 7784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.507239164130501e-06,
+      "loss": 0.8888,
+      "step": 7785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5047870164305576e-06,
+      "loss": 0.8431,
+      "step": 7786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.502335207433136e-06,
+      "loss": 0.9014,
+      "step": 7787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.49988373732296e-06,
+      "loss": 0.8944,
+      "step": 7788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497432606284758e-06,
+      "loss": 0.8703,
+      "step": 7789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494981814503199e-06,
+      "loss": 0.8678,
+      "step": 7790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4925313621629585e-06,
+      "loss": 0.8797,
+      "step": 7791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4900812494486575e-06,
+      "loss": 0.8703,
+      "step": 7792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.487631476544921e-06,
+      "loss": 0.9237,
+      "step": 7793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48518204363632e-06,
+      "loss": 0.8973,
+      "step": 7794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.482732950907419e-06,
+      "loss": 0.9124,
+      "step": 7795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480284198542749e-06,
+      "loss": 0.8855,
+      "step": 7796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477835786726817e-06,
+      "loss": 0.8992,
+      "step": 7797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.475387715644107e-06,
+      "loss": 0.8591,
+      "step": 7798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.472939985479071e-06,
+      "loss": 0.8894,
+      "step": 7799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4704925964161416e-06,
+      "loss": 0.8611,
+      "step": 7800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468045548639722e-06,
+      "loss": 0.8373,
+      "step": 7801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.465598842334192e-06,
+      "loss": 0.8701,
+      "step": 7802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.463152477683902e-06,
+      "loss": 0.8519,
+      "step": 7803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4607064548731825e-06,
+      "loss": 0.8821,
+      "step": 7804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.458260774086332e-06,
+      "loss": 0.8951,
+      "step": 7805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.455815435507627e-06,
+      "loss": 0.8603,
+      "step": 7806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4533704393213204e-06,
+      "loss": 0.8634,
+      "step": 7807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450925785711632e-06,
+      "loss": 0.8665,
+      "step": 7808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4484814748627614e-06,
+      "loss": 0.8681,
+      "step": 7809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4460375069588815e-06,
+      "loss": 0.9224,
+      "step": 7810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.443593882184139e-06,
+      "loss": 0.8838,
+      "step": 7811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.441150600722653e-06,
+      "loss": 0.8731,
+      "step": 7812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.438707662758521e-06,
+      "loss": 0.8951,
+      "step": 7813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.436265068475815e-06,
+      "loss": 0.9006,
+      "step": 7814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433822818058565e-06,
+      "loss": 0.903,
+      "step": 7815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431380911690805e-06,
+      "loss": 0.8791,
+      "step": 7816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4289393495565076e-06,
+      "loss": 0.8588,
+      "step": 7817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426498131839659e-06,
+      "loss": 0.9247,
+      "step": 7818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.42405725872418e-06,
+      "loss": 0.5713,
+      "step": 7819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421616730394e-06,
+      "loss": 0.8287,
+      "step": 7820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.419176547032991e-06,
+      "loss": 0.8557,
+      "step": 7821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416736708825025e-06,
+      "loss": 0.4842,
+      "step": 7822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414297215953937e-06,
+      "loss": 0.9097,
+      "step": 7823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.41185806860353e-06,
+      "loss": 0.5957,
+      "step": 7824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4094192669575975e-06,
+      "loss": 0.8752,
+      "step": 7825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406980811199881e-06,
+      "loss": 0.8391,
+      "step": 7826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404542701514131e-06,
+      "loss": 0.5424,
+      "step": 7827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402104938084034e-06,
+      "loss": 0.8698,
+      "step": 7828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399667521093285e-06,
+      "loss": 0.8906,
+      "step": 7829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397230450725523e-06,
+      "loss": 0.8394,
+      "step": 7830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.394793727164389e-06,
+      "loss": 0.9103,
+      "step": 7831
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.392357350593469e-06,
+      "loss": 0.7983,
+      "step": 7832
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3899213211963545e-06,
+      "loss": 0.8961,
+      "step": 7833
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.387485639156575e-06,
+      "loss": 0.8311,
+      "step": 7834
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.38505030465767e-06,
+      "loss": 0.8413,
+      "step": 7835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3826153178831245e-06,
+      "loss": 0.8843,
+      "step": 7836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.380180679016411e-06,
+      "loss": 0.847,
+      "step": 7837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3777463882409744e-06,
+      "loss": 0.8642,
+      "step": 7838
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.375312445740232e-06,
+      "loss": 0.8885,
+      "step": 7839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3728788516975734e-06,
+      "loss": 0.8945,
+      "step": 7840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3704456062963674e-06,
+      "loss": 0.8936,
+      "step": 7841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.368012709719948e-06,
+      "loss": 0.9278,
+      "step": 7842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3655801621516305e-06,
+      "loss": 0.8848,
+      "step": 7843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3631479637747e-06,
+      "loss": 0.8729,
+      "step": 7844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360716114772415e-06,
+      "loss": 0.9053,
+      "step": 7845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.358284615328011e-06,
+      "loss": 0.8326,
+      "step": 7846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.355853465624695e-06,
+      "loss": 0.8435,
+      "step": 7847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.353422665845647e-06,
+      "loss": 0.902,
+      "step": 7848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.350992216174021e-06,
+      "loss": 0.8804,
+      "step": 7849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.348562116792946e-06,
+      "loss": 0.8447,
+      "step": 7850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3461323678855235e-06,
+      "loss": 0.8776,
+      "step": 7851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.343702969634828e-06,
+      "loss": 0.8346,
+      "step": 7852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.341273922223908e-06,
+      "loss": 0.9044,
+      "step": 7853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.338845225835788e-06,
+      "loss": 0.907,
+      "step": 7854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.336416880653461e-06,
+      "loss": 0.884,
+      "step": 7855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.333988886859903e-06,
+      "loss": 0.9047,
+      "step": 7856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.331561244638044e-06,
+      "loss": 0.8185,
+      "step": 7857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.329133954170816e-06,
+      "loss": 0.9111,
+      "step": 7858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.326707015641093e-06,
+      "loss": 0.8779,
+      "step": 7859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324280429231754e-06,
+      "loss": 0.8913,
+      "step": 7860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3218541951256215e-06,
+      "loss": 0.9165,
+      "step": 7861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.319428313505523e-06,
+      "loss": 0.903,
+      "step": 7862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.317002784554221e-06,
+      "loss": 0.9093,
+      "step": 7863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.314577608454494e-06,
+      "loss": 0.8835,
+      "step": 7864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.312152785389056e-06,
+      "loss": 0.9232,
+      "step": 7865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30972831554062e-06,
+      "loss": 0.8738,
+      "step": 7866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307304199091865e-06,
+      "loss": 0.8517,
+      "step": 7867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.304880436225432e-06,
+      "loss": 0.9307,
+      "step": 7868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302457027123957e-06,
+      "loss": 0.8862,
+      "step": 7869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300033971970027e-06,
+      "loss": 0.891,
+      "step": 7870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.297611270946223e-06,
+      "loss": 0.8952,
+      "step": 7871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.295188924235078e-06,
+      "loss": 0.8903,
+      "step": 7872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292766932019121e-06,
+      "loss": 0.7863,
+      "step": 7873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2903452944808294e-06,
+      "loss": 0.9085,
+      "step": 7874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.287924011802685e-06,
+      "loss": 0.8659,
+      "step": 7875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2855030841671054e-06,
+      "loss": 0.8345,
+      "step": 7876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283082511756519e-06,
+      "loss": 0.5233,
+      "step": 7877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2806622947532985e-06,
+      "loss": 0.8732,
+      "step": 7878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.278242433339802e-06,
+      "loss": 0.5668,
+      "step": 7879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.275822927698362e-06,
+      "loss": 0.8627,
+      "step": 7880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.273403778011283e-06,
+      "loss": 0.8933,
+      "step": 7881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.270984984460838e-06,
+      "loss": 0.9038,
+      "step": 7882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2685665472292805e-06,
+      "loss": 0.8471,
+      "step": 7883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.26614846649883e-06,
+      "loss": 0.8589,
+      "step": 7884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263730742451685e-06,
+      "loss": 0.9094,
+      "step": 7885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2613133752700145e-06,
+      "loss": 0.8672,
+      "step": 7886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.258896365135959e-06,
+      "loss": 0.901,
+      "step": 7887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256479712231637e-06,
+      "loss": 0.8652,
+      "step": 7888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2540634167391325e-06,
+      "loss": 0.8951,
+      "step": 7889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.251647478840511e-06,
+      "loss": 0.8834,
+      "step": 7890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249231898717806e-06,
+      "loss": 0.8772,
+      "step": 7891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.246816676553024e-06,
+      "loss": 0.8457,
+      "step": 7892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244401812528146e-06,
+      "loss": 0.8434,
+      "step": 7893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.241987306825127e-06,
+      "loss": 0.4934,
+      "step": 7894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2395731596258925e-06,
+      "loss": 0.8659,
+      "step": 7895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.237159371112342e-06,
+      "loss": 0.8272,
+      "step": 7896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234745941466348e-06,
+      "loss": 0.925,
+      "step": 7897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.232332870869763e-06,
+      "loss": 0.8628,
+      "step": 7898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.229920159504389e-06,
+      "loss": 0.8521,
+      "step": 7899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.227507807552037e-06,
+      "loss": 0.9214,
+      "step": 7900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225095815194453e-06,
+      "loss": 0.876,
+      "step": 7901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.222684182613392e-06,
+      "loss": 0.8011,
+      "step": 7902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.220272909990548e-06,
+      "loss": 0.8654,
+      "step": 7903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217861997507618e-06,
+      "loss": 0.906,
+      "step": 7904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.215451445346245e-06,
+      "loss": 0.8733,
+      "step": 7905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.213041253688074e-06,
+      "loss": 0.8551,
+      "step": 7906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.210631422714686e-06,
+      "loss": 0.8924,
+      "step": 7907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2082219526076745e-06,
+      "loss": 0.8824,
+      "step": 7908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.205812843548577e-06,
+      "loss": 0.8722,
+      "step": 7909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20340409571891e-06,
+      "loss": 0.8938,
+      "step": 7910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200995709300179e-06,
+      "loss": 0.8652,
+      "step": 7911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198587684473836e-06,
+      "loss": 0.8991,
+      "step": 7912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.196180021421332e-06,
+      "loss": 0.8988,
+      "step": 7913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.193772720324064e-06,
+      "loss": 0.8621,
+      "step": 7914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.191365781363431e-06,
+      "loss": 0.8384,
+      "step": 7915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188959204720776e-06,
+      "loss": 0.8734,
+      "step": 7916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1865529905774405e-06,
+      "loss": 0.8812,
+      "step": 7917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184147139114713e-06,
+      "loss": 0.8748,
+      "step": 7918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.181741650513883e-06,
+      "loss": 0.8441,
+      "step": 7919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.179336524956185e-06,
+      "loss": 0.9264,
+      "step": 7920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1769317626228455e-06,
+      "loss": 0.8631,
+      "step": 7921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1745273636950545e-06,
+      "loss": 0.8746,
+      "step": 7922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.172123328353978e-06,
+      "loss": 0.8695,
+      "step": 7923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.169719656780754e-06,
+      "loss": 0.9327,
+      "step": 7924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.167316349156495e-06,
+      "loss": 0.8589,
+      "step": 7925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1649134056622795e-06,
+      "loss": 0.863,
+      "step": 7926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162510826479166e-06,
+      "loss": 0.9095,
+      "step": 7927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16010861178818e-06,
+      "loss": 0.888,
+      "step": 7928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157706761770326e-06,
+      "loss": 0.8346,
+      "step": 7929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1553052766065735e-06,
+      "loss": 0.8636,
+      "step": 7930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15290415647787e-06,
+      "loss": 0.8517,
+      "step": 7931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1505034015651345e-06,
+      "loss": 0.851,
+      "step": 7932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148103012049255e-06,
+      "loss": 0.4879,
+      "step": 7933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.145702988111095e-06,
+      "loss": 0.8794,
+      "step": 7934
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1433033299314915e-06,
+      "loss": 0.8999,
+      "step": 7935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.140904037691252e-06,
+      "loss": 0.8262,
+      "step": 7936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138505111571157e-06,
+      "loss": 0.8638,
+      "step": 7937
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136106551751958e-06,
+      "loss": 0.8765,
+      "step": 7938
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133708358414382e-06,
+      "loss": 0.8673,
+      "step": 7939
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.131310531739129e-06,
+      "loss": 0.8621,
+      "step": 7940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.128913071906858e-06,
+      "loss": 0.8368,
+      "step": 7941
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1265159790982275e-06,
+      "loss": 0.875,
+      "step": 7942
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1241192534938355e-06,
+      "loss": 0.8453,
+      "step": 7943
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.121722895274284e-06,
+      "loss": 0.8863,
+      "step": 7944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119326904620119e-06,
+      "loss": 0.9036,
+      "step": 7945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.116931281711886e-06,
+      "loss": 0.898,
+      "step": 7946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.114536026730073e-06,
+      "loss": 0.8445,
+      "step": 7947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112141139855171e-06,
+      "loss": 0.8759,
+      "step": 7948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1097466212676175e-06,
+      "loss": 0.8615,
+      "step": 7949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.107352471147844e-06,
+      "loss": 0.8624,
+      "step": 7950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.104958689676234e-06,
+      "loss": 0.8887,
+      "step": 7951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.102565277033155e-06,
+      "loss": 0.872,
+      "step": 7952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.100172233398946e-06,
+      "loss": 0.8967,
+      "step": 7953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.09777955895391e-06,
+      "loss": 0.8944,
+      "step": 7954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.095387253878346e-06,
+      "loss": 0.8602,
+      "step": 7955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092995318352485e-06,
+      "loss": 0.8914,
+      "step": 7956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090603752556575e-06,
+      "loss": 0.8689,
+      "step": 7957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.088212556670795e-06,
+      "loss": 0.881,
+      "step": 7958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.085821730875331e-06,
+      "loss": 0.83,
+      "step": 7959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083431275350312e-06,
+      "loss": 0.8899,
+      "step": 7960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0810411902758675e-06,
+      "loss": 0.8413,
+      "step": 7961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0786514758320706e-06,
+      "loss": 0.9088,
+      "step": 7962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.076262132198986e-06,
+      "loss": 0.8421,
+      "step": 7963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.073873159556643e-06,
+      "loss": 0.8438,
+      "step": 7964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.071484558085044e-06,
+      "loss": 0.8616,
+      "step": 7965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.069096327964167e-06,
+      "loss": 0.4821,
+      "step": 7966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.066708469373958e-06,
+      "loss": 0.8807,
+      "step": 7967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.064320982494332e-06,
+      "loss": 0.8993,
+      "step": 7968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.061933867505185e-06,
+      "loss": 0.8781,
+      "step": 7969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0595471245863745e-06,
+      "loss": 0.8382,
+      "step": 7970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.057160753917741e-06,
+      "loss": 0.8595,
+      "step": 7971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.054774755679086e-06,
+      "loss": 0.9037,
+      "step": 7972
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.052389130050193e-06,
+      "loss": 0.8921,
+      "step": 7973
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.05000387721081e-06,
+      "loss": 0.8757,
+      "step": 7974
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.047618997340659e-06,
+      "loss": 0.906,
+      "step": 7975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045234490619435e-06,
+      "loss": 0.8531,
+      "step": 7976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.042850357226807e-06,
+      "loss": 0.8608,
+      "step": 7977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.04046659734241e-06,
+      "loss": 0.866,
+      "step": 7978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038083211145854e-06,
+      "loss": 0.8402,
+      "step": 7979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.035700198816722e-06,
+      "loss": 0.9445,
+      "step": 7980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.033317560534568e-06,
+      "loss": 0.918,
+      "step": 7981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.030935296478922e-06,
+      "loss": 0.9068,
+      "step": 7982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.028553406829268e-06,
+      "loss": 0.8512,
+      "step": 7983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.026171891765092e-06,
+      "loss": 0.8694,
+      "step": 7984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.023790751465818e-06,
+      "loss": 0.8894,
+      "step": 7985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021409986110875e-06,
+      "loss": 0.8395,
+      "step": 7986
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019029595879633e-06,
+      "loss": 0.8745,
+      "step": 7987
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.016649580951462e-06,
+      "loss": 0.8808,
+      "step": 7988
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014269941505675e-06,
+      "loss": 0.8455,
+      "step": 7989
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.011890677721587e-06,
+      "loss": 0.8589,
+      "step": 7990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.009511789778454e-06,
+      "loss": 0.8677,
+      "step": 7991
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007133277855534e-06,
+      "loss": 0.7975,
+      "step": 7992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.004755142132032e-06,
+      "loss": 0.795,
+      "step": 7993
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.002377382787135e-06,
+      "loss": 0.8575,
+      "step": 7994
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.8678,
+      "step": 7995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.997622993949765e-06,
+      "loss": 0.8665,
+      "step": 7996
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.995246364815522e-06,
+      "loss": 0.8843,
+      "step": 7997
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.992870112776342e-06,
+      "loss": 0.8554,
+      "step": 7998
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9904942380112846e-06,
+      "loss": 0.5764,
+      "step": 7999
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9881187406993455e-06,
+      "loss": 0.8636,
+      "step": 8000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.985743621019531e-06,
+      "loss": 0.8595,
+      "step": 8001
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.983368879150787e-06,
+      "loss": 0.8402,
+      "step": 8002
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.98099451527205e-06,
+      "loss": 0.8755,
+      "step": 8003
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9786205295622184e-06,
+      "loss": 0.8279,
+      "step": 8004
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.976246922200169e-06,
+      "loss": 0.8787,
+      "step": 8005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973873693364746e-06,
+      "loss": 0.851,
+      "step": 8006
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.971500843234765e-06,
+      "loss": 0.8853,
+      "step": 8007
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969128371989015e-06,
+      "loss": 0.8651,
+      "step": 8008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.966756279806255e-06,
+      "loss": 0.8712,
+      "step": 8009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.964384566865217e-06,
+      "loss": 0.8532,
+      "step": 8010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.962013233344603e-06,
+      "loss": 0.8806,
+      "step": 8011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959642279423085e-06,
+      "loss": 0.8431,
+      "step": 8012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957271705279309e-06,
+      "loss": 0.9089,
+      "step": 8013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9549015110918945e-06,
+      "loss": 0.876,
+      "step": 8014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.952531697039424e-06,
+      "loss": 0.8416,
+      "step": 8015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.950162263300462e-06,
+      "loss": 0.9239,
+      "step": 8016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947793210053537e-06,
+      "loss": 0.8657,
+      "step": 8017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.945424537477149e-06,
+      "loss": 0.8829,
+      "step": 8018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.943056245749775e-06,
+      "loss": 0.8599,
+      "step": 8019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940688335049857e-06,
+      "loss": 0.8758,
+      "step": 8020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938320805555811e-06,
+      "loss": 0.8437,
+      "step": 8021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9359536574460244e-06,
+      "loss": 0.8334,
+      "step": 8022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933586890898857e-06,
+      "loss": 0.9043,
+      "step": 8023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931220506092641e-06,
+      "loss": 0.8474,
+      "step": 8024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928854503205666e-06,
+      "loss": 0.8677,
+      "step": 8025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.926488882416218e-06,
+      "loss": 0.8954,
+      "step": 8026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9241236439025275e-06,
+      "loss": 0.8981,
+      "step": 8027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921758787842823e-06,
+      "loss": 0.8835,
+      "step": 8028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.919394314415276e-06,
+      "loss": 0.8916,
+      "step": 8029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917030223798057e-06,
+      "loss": 0.9052,
+      "step": 8030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914666516169279e-06,
+      "loss": 0.8431,
+      "step": 8031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912303191707059e-06,
+      "loss": 0.8559,
+      "step": 8032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909940250589448e-06,
+      "loss": 0.8516,
+      "step": 8033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.907577692994505e-06,
+      "loss": 0.8874,
+      "step": 8034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.905215519100231e-06,
+      "loss": 0.8681,
+      "step": 8035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.902853729084615e-06,
+      "loss": 0.8486,
+      "step": 8036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9004923231256075e-06,
+      "loss": 0.8499,
+      "step": 8037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8981313014011375e-06,
+      "loss": 0.8452,
+      "step": 8038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895770664089101e-06,
+      "loss": 0.8721,
+      "step": 8039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.893410411367366e-06,
+      "loss": 0.8785,
+      "step": 8040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891050543413771e-06,
+      "loss": 0.8557,
+      "step": 8041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.888691060406122e-06,
+      "loss": 0.8838,
+      "step": 8042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8863319625222126e-06,
+      "loss": 0.8447,
+      "step": 8043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8839732499397805e-06,
+      "loss": 0.877,
+      "step": 8044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881614922836555e-06,
+      "loss": 0.8433,
+      "step": 8045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8792569813902295e-06,
+      "loss": 0.8794,
+      "step": 8046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876899425778466e-06,
+      "loss": 0.8757,
+      "step": 8047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874542256178903e-06,
+      "loss": 0.8783,
+      "step": 8048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.872185472769146e-06,
+      "loss": 0.8296,
+      "step": 8049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.869829075726774e-06,
+      "loss": 0.8975,
+      "step": 8050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.867473065229332e-06,
+      "loss": 0.8893,
+      "step": 8051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.865117441454342e-06,
+      "loss": 0.5837,
+      "step": 8052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.862762204579294e-06,
+      "loss": 0.9059,
+      "step": 8053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860407354781647e-06,
+      "loss": 0.8154,
+      "step": 8054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8580528922388395e-06,
+      "loss": 0.8826,
+      "step": 8055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85569881712826e-06,
+      "loss": 0.873,
+      "step": 8056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8533451296272934e-06,
+      "loss": 0.8711,
+      "step": 8057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.850991829913282e-06,
+      "loss": 0.874,
+      "step": 8058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84863891816354e-06,
+      "loss": 0.8519,
+      "step": 8059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.846286394555352e-06,
+      "loss": 0.8342,
+      "step": 8060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8439342592659745e-06,
+      "loss": 0.8865,
+      "step": 8061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.841582512472636e-06,
+      "loss": 0.8508,
+      "step": 8062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.839231154352535e-06,
+      "loss": 0.8679,
+      "step": 8063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836880185082839e-06,
+      "loss": 0.842,
+      "step": 8064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.834529604840686e-06,
+      "loss": 0.9225,
+      "step": 8065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.832179413803193e-06,
+      "loss": 0.5141,
+      "step": 8066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.829829612147427e-06,
+      "loss": 0.9166,
+      "step": 8067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.827480200050456e-06,
+      "loss": 0.8519,
+      "step": 8068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.825131177689286e-06,
+      "loss": 0.9056,
+      "step": 8069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822782545240925e-06,
+      "loss": 0.8628,
+      "step": 8070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820434302882321e-06,
+      "loss": 0.9127,
+      "step": 8071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818086450790423e-06,
+      "loss": 0.8519,
+      "step": 8072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8157389891421205e-06,
+      "loss": 0.8237,
+      "step": 8073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.813391918114304e-06,
+      "loss": 0.886,
+      "step": 8074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811045237883803e-06,
+      "loss": 0.803,
+      "step": 8075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808698948627452e-06,
+      "loss": 0.8448,
+      "step": 8076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806353050522021e-06,
+      "loss": 0.8705,
+      "step": 8077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.804007543744277e-06,
+      "loss": 0.8669,
+      "step": 8078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801662428470943e-06,
+      "loss": 0.8636,
+      "step": 8079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.79931770487872e-06,
+      "loss": 0.8497,
+      "step": 8080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.796973373144276e-06,
+      "loss": 0.8662,
+      "step": 8081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.794629433444249e-06,
+      "loss": 0.9048,
+      "step": 8082
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.792285885955251e-06,
+      "loss": 0.8108,
+      "step": 8083
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.78994273085386e-06,
+      "loss": 0.8611,
+      "step": 8084
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.78759996831663e-06,
+      "loss": 0.8902,
+      "step": 8085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.785257598520078e-06,
+      "loss": 0.5275,
+      "step": 8086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.782915621640697e-06,
+      "loss": 0.898,
+      "step": 8087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7805740378549495e-06,
+      "loss": 0.8673,
+      "step": 8088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.778232847339267e-06,
+      "loss": 0.8609,
+      "step": 8089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.775892050270051e-06,
+      "loss": 0.9398,
+      "step": 8090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.773551646823677e-06,
+      "loss": 0.9116,
+      "step": 8091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.771211637176487e-06,
+      "loss": 0.8228,
+      "step": 8092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.768872021504795e-06,
+      "loss": 0.8829,
+      "step": 8093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.766532799984883e-06,
+      "loss": 0.9041,
+      "step": 8094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7641939727930064e-06,
+      "loss": 0.8907,
+      "step": 8095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.761855540105391e-06,
+      "loss": 0.8472,
+      "step": 8096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.759517502098234e-06,
+      "loss": 0.934,
+      "step": 8097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757179858947689e-06,
+      "loss": 0.5297,
+      "step": 8098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754842610829908e-06,
+      "loss": 0.8948,
+      "step": 8099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7525057579209775e-06,
+      "loss": 0.8642,
+      "step": 8100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.75016930039699e-06,
+      "loss": 0.8958,
+      "step": 8101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7478332384339834e-06,
+      "loss": 0.843,
+      "step": 8102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.745497572207975e-06,
+      "loss": 0.8397,
+      "step": 8103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743162301894952e-06,
+      "loss": 0.8937,
+      "step": 8104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.740827427670871e-06,
+      "loss": 0.8373,
+      "step": 8105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.738492949711659e-06,
+      "loss": 0.8792,
+      "step": 8106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.736158868193209e-06,
+      "loss": 0.8488,
+      "step": 8107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.733825183291396e-06,
+      "loss": 0.9054,
+      "step": 8108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731491895182045e-06,
+      "loss": 0.4708,
+      "step": 8109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.729159004040976e-06,
+      "loss": 0.894,
+      "step": 8110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726826510043953e-06,
+      "loss": 0.8721,
+      "step": 8111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724494413366739e-06,
+      "loss": 0.8503,
+      "step": 8112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722162714185035e-06,
+      "loss": 0.8482,
+      "step": 8113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7198314126745424e-06,
+      "loss": 0.8864,
+      "step": 8114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7175005090109074e-06,
+      "loss": 0.8715,
+      "step": 8115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715170003369768e-06,
+      "loss": 0.8735,
+      "step": 8116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7128398959267095e-06,
+      "loss": 0.8607,
+      "step": 8117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.710510186857313e-06,
+      "loss": 0.8453,
+      "step": 8118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.708180876337107e-06,
+      "loss": 0.8862,
+      "step": 8119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7058519645416004e-06,
+      "loss": 0.8872,
+      "step": 8120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.703523451646271e-06,
+      "loss": 0.8727,
+      "step": 8121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.701195337826566e-06,
+      "loss": 0.8903,
+      "step": 8122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.698867623257905e-06,
+      "loss": 0.4842,
+      "step": 8123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696540308115672e-06,
+      "loss": 0.8589,
+      "step": 8124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696540308115672e-06,
+      "loss": 0.9058,
+      "step": 8125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6942133925752245e-06,
+      "loss": 0.849,
+      "step": 8126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6918868768118906e-06,
+      "loss": 0.8786,
+      "step": 8127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689560761000968e-06,
+      "loss": 0.8498,
+      "step": 8128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.687235045317723e-06,
+      "loss": 0.8328,
+      "step": 8129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68490972993739e-06,
+      "loss": 0.9043,
+      "step": 8130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.682584815035177e-06,
+      "loss": 0.8394,
+      "step": 8131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680260300786261e-06,
+      "loss": 0.5151,
+      "step": 8132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677936187365787e-06,
+      "loss": 0.864,
+      "step": 8133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.675612474948872e-06,
+      "loss": 0.8866,
+      "step": 8134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.673289163710599e-06,
+      "loss": 0.8867,
+      "step": 8135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.670966253826027e-06,
+      "loss": 0.8147,
+      "step": 8136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668643745470178e-06,
+      "loss": 0.8863,
+      "step": 8137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.666321638818049e-06,
+      "loss": 0.9045,
+      "step": 8138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6639999340446045e-06,
+      "loss": 0.8521,
+      "step": 8139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661678631324781e-06,
+      "loss": 0.8821,
+      "step": 8140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.659357730833472e-06,
+      "loss": 0.9035,
+      "step": 8141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6570372327455686e-06,
+      "loss": 0.8943,
+      "step": 8142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654717137235895e-06,
+      "loss": 0.8937,
+      "step": 8143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.652397444479283e-06,
+      "loss": 0.8752,
+      "step": 8144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.650078154650498e-06,
+      "loss": 0.8632,
+      "step": 8145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647759267924306e-06,
+      "loss": 0.8626,
+      "step": 8146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645440784475422e-06,
+      "loss": 0.8761,
+      "step": 8147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.643122704478541e-06,
+      "loss": 0.8174,
+      "step": 8148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6408050281083225e-06,
+      "loss": 0.8344,
+      "step": 8149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6384877555393985e-06,
+      "loss": 0.8098,
+      "step": 8150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.636170886946371e-06,
+      "loss": 0.8514,
+      "step": 8151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.633854422503801e-06,
+      "loss": 0.8814,
+      "step": 8152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.631538362386241e-06,
+      "loss": 0.8945,
+      "step": 8153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629222706768186e-06,
+      "loss": 0.8493,
+      "step": 8154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626907455824131e-06,
+      "loss": 0.8817,
+      "step": 8155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624592609728508e-06,
+      "loss": 0.8969,
+      "step": 8156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6222781686557485e-06,
+      "loss": 0.89,
+      "step": 8157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619964132780225e-06,
+      "loss": 0.8689,
+      "step": 8158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.61765050227631e-06,
+      "loss": 0.8593,
+      "step": 8159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.615337277318313e-06,
+      "loss": 0.8638,
+      "step": 8160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613024458080547e-06,
+      "loss": 0.8151,
+      "step": 8161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610712044737264e-06,
+      "loss": 0.9349,
+      "step": 8162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.608400037462702e-06,
+      "loss": 0.8292,
+      "step": 8163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606088436431064e-06,
+      "loss": 0.8628,
+      "step": 8164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6037772418165235e-06,
+      "loss": 0.9165,
+      "step": 8165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.601466453793224e-06,
+      "loss": 0.89,
+      "step": 8166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599156072535277e-06,
+      "loss": 0.8853,
+      "step": 8167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596846098216763e-06,
+      "loss": 0.8611,
+      "step": 8168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5945365310117325e-06,
+      "loss": 0.872,
+      "step": 8169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592227371094206e-06,
+      "loss": 0.8486,
+      "step": 8170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.589918618638173e-06,
+      "loss": 0.8571,
+      "step": 8171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5876102738175895e-06,
+      "loss": 0.9153,
+      "step": 8172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585302336806386e-06,
+      "loss": 0.8351,
+      "step": 8173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.582994807778458e-06,
+      "loss": 0.9068,
+      "step": 8174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.580687686907673e-06,
+      "loss": 0.8944,
+      "step": 8175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5783809743678654e-06,
+      "loss": 0.472,
+      "step": 8176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57607467033284e-06,
+      "loss": 0.5334,
+      "step": 8177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573768774976371e-06,
+      "loss": 0.8414,
+      "step": 8178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571463288472202e-06,
+      "loss": 0.857,
+      "step": 8179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569158210994045e-06,
+      "loss": 0.8373,
+      "step": 8180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5668535427155816e-06,
+      "loss": 0.8742,
+      "step": 8181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.564549283810465e-06,
+      "loss": 0.8638,
+      "step": 8182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.562245434452306e-06,
+      "loss": 0.8695,
+      "step": 8183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5599419948147075e-06,
+      "loss": 0.8999,
+      "step": 8184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.557638965071213e-06,
+      "loss": 0.8304,
+      "step": 8185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.555336345395365e-06,
+      "loss": 0.8645,
+      "step": 8186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.553034135960643e-06,
+      "loss": 0.8408,
+      "step": 8187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.55073233694053e-06,
+      "loss": 0.9066,
+      "step": 8188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548430948508443e-06,
+      "loss": 0.8327,
+      "step": 8189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546129970837799e-06,
+      "loss": 0.836,
+      "step": 8190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543829404101966e-06,
+      "loss": 0.8996,
+      "step": 8191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.541529248474286e-06,
+      "loss": 0.8771,
+      "step": 8192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.539229504128073e-06,
+      "loss": 0.8889,
+      "step": 8193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536930171236596e-06,
+      "loss": 0.8731,
+      "step": 8194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534631249973118e-06,
+      "loss": 0.872,
+      "step": 8195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532332740510842e-06,
+      "loss": 0.8364,
+      "step": 8196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530034643022969e-06,
+      "loss": 0.8804,
+      "step": 8197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527736957682642e-06,
+      "loss": 0.8412,
+      "step": 8198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5254396846629975e-06,
+      "loss": 0.8457,
+      "step": 8199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5231428241371154e-06,
+      "loss": 0.914,
+      "step": 8200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.520846376278074e-06,
+      "loss": 0.8662,
+      "step": 8201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.518550341258888e-06,
+      "loss": 0.8203,
+      "step": 8202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.516254719252574e-06,
+      "loss": 0.9268,
+      "step": 8203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.513959510432089e-06,
+      "loss": 0.8408,
+      "step": 8204
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.511664714970374e-06,
+      "loss": 0.8559,
+      "step": 8205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5093703330403385e-06,
+      "loss": 0.845,
+      "step": 8206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.507076364814855e-06,
+      "loss": 0.889,
+      "step": 8207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.50478281046677e-06,
+      "loss": 0.8993,
+      "step": 8208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5024896701688955e-06,
+      "loss": 0.8708,
+      "step": 8209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.500196944094015e-06,
+      "loss": 0.8709,
+      "step": 8210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.497904632414879e-06,
+      "loss": 0.8731,
+      "step": 8211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.495612735304207e-06,
+      "loss": 0.8715,
+      "step": 8212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493321252934686e-06,
+      "loss": 0.8423,
+      "step": 8213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491030185478976e-06,
+      "loss": 0.8706,
+      "step": 8214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488739533109701e-06,
+      "loss": 0.8395,
+      "step": 8215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.486449295999455e-06,
+      "loss": 0.8837,
+      "step": 8216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.484159474320804e-06,
+      "loss": 0.8863,
+      "step": 8217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481870068246278e-06,
+      "loss": 0.8982,
+      "step": 8218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.479581077948378e-06,
+      "loss": 0.9559,
+      "step": 8219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477292503599574e-06,
+      "loss": 0.8856,
+      "step": 8220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475004345372304e-06,
+      "loss": 0.8659,
+      "step": 8221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4727166034389746e-06,
+      "loss": 0.8931,
+      "step": 8222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470429277971961e-06,
+      "loss": 0.8303,
+      "step": 8223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46814236914361e-06,
+      "loss": 0.8623,
+      "step": 8224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.465855877126225e-06,
+      "loss": 0.534,
+      "step": 8225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4635698020921016e-06,
+      "loss": 0.9011,
+      "step": 8226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.461284144213473e-06,
+      "loss": 0.8695,
+      "step": 8227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.458998903662574e-06,
+      "loss": 0.8578,
+      "step": 8228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.456714080611575e-06,
+      "loss": 0.8747,
+      "step": 8229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454429675232649e-06,
+      "loss": 0.8782,
+      "step": 8230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.452145687697903e-06,
+      "loss": 0.8532,
+      "step": 8231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449862118179444e-06,
+      "loss": 0.8963,
+      "step": 8232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449862118179444e-06,
+      "loss": 0.9382,
+      "step": 8233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4475789668493205e-06,
+      "loss": 0.8804,
+      "step": 8234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4452962338795715e-06,
+      "loss": 0.8267,
+      "step": 8235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443013919442194e-06,
+      "loss": 0.8439,
+      "step": 8236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.440732023709145e-06,
+      "loss": 0.8858,
+      "step": 8237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.438450546852373e-06,
+      "loss": 0.8768,
+      "step": 8238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436169489043768e-06,
+      "loss": 0.5029,
+      "step": 8239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.433888850455215e-06,
+      "loss": 0.8877,
+      "step": 8240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4316086312585396e-06,
+      "loss": 0.8734,
+      "step": 8241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.429328831625565e-06,
+      "loss": 0.8446,
+      "step": 8242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4270494517280545e-06,
+      "loss": 0.829,
+      "step": 8243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4247704917377676e-06,
+      "loss": 0.8773,
+      "step": 8244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422491951826402e-06,
+      "loss": 0.8938,
+      "step": 8245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420213832165657e-06,
+      "loss": 0.8771,
+      "step": 8246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.417936132927169e-06,
+      "loss": 0.8819,
+      "step": 8247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4156588542825625e-06,
+      "loss": 0.8716,
+      "step": 8248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.413381996403423e-06,
+      "loss": 0.8932,
+      "step": 8249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.411105559461307e-06,
+      "loss": 0.8685,
+      "step": 8250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408829543627737e-06,
+      "loss": 0.5033,
+      "step": 8251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406553949074207e-06,
+      "loss": 0.8632,
+      "step": 8252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.404278775972174e-06,
+      "loss": 0.8182,
+      "step": 8253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402004024493069e-06,
+      "loss": 0.8799,
+      "step": 8254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399729694808285e-06,
+      "loss": 0.8611,
+      "step": 8255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397455787089191e-06,
+      "loss": 0.8928,
+      "step": 8256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3951823015071186e-06,
+      "loss": 0.8797,
+      "step": 8257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3929092382333675e-06,
+      "loss": 0.918,
+      "step": 8258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.390636597439207e-06,
+      "loss": 0.9025,
+      "step": 8259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.388364379295878e-06,
+      "loss": 0.848,
+      "step": 8260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386092583974582e-06,
+      "loss": 0.8226,
+      "step": 8261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383821211646495e-06,
+      "loss": 0.8499,
+      "step": 8262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381550262482759e-06,
+      "loss": 0.8911,
+      "step": 8263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379279736654484e-06,
+      "loss": 0.8713,
+      "step": 8264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377009634332747e-06,
+      "loss": 0.85,
+      "step": 8265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.374739955688595e-06,
+      "loss": 0.8909,
+      "step": 8266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372470700893046e-06,
+      "loss": 0.8941,
+      "step": 8267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370201870117074e-06,
+      "loss": 0.5685,
+      "step": 8268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3679334635316395e-06,
+      "loss": 0.8488,
+      "step": 8269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365665481307648e-06,
+      "loss": 0.8854,
+      "step": 8270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363397923616003e-06,
+      "loss": 0.8341,
+      "step": 8271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361130790627541e-06,
+      "loss": 0.8439,
+      "step": 8272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.358864082513101e-06,
+      "loss": 0.8731,
+      "step": 8273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356597799443459e-06,
+      "loss": 0.8462,
+      "step": 8274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.354331941589387e-06,
+      "loss": 0.8415,
+      "step": 8275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.352066509121599e-06,
+      "loss": 0.8561,
+      "step": 8276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.349801502210801e-06,
+      "loss": 0.8687,
+      "step": 8277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347536921027646e-06,
+      "loss": 0.8526,
+      "step": 8278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345272765742765e-06,
+      "loss": 0.8706,
+      "step": 8279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.343009036526766e-06,
+      "loss": 0.8331,
+      "step": 8280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3407457335502e-06,
+      "loss": 0.8423,
+      "step": 8281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.338482856983617e-06,
+      "loss": 0.8615,
+      "step": 8282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.336220406997503e-06,
+      "loss": 0.8956,
+      "step": 8283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333958383762345e-06,
+      "loss": 0.8868,
+      "step": 8284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3316967874485605e-06,
+      "loss": 0.8493,
+      "step": 8285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.329435618226574e-06,
+      "loss": 0.8677,
+      "step": 8286
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327174876266743e-06,
+      "loss": 0.86,
+      "step": 8287
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324914561739423e-06,
+      "loss": 0.8912,
+      "step": 8288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322654674814912e-06,
+      "loss": 0.8971,
+      "step": 8289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.320395215663488e-06,
+      "loss": 0.862,
+      "step": 8290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318136184455398e-06,
+      "loss": 0.8446,
+      "step": 8291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315877581360853e-06,
+      "loss": 0.9019,
+      "step": 8292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313619406550034e-06,
+      "loss": 0.8888,
+      "step": 8293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311361660193087e-06,
+      "loss": 0.9001,
+      "step": 8294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3091043424601285e-06,
+      "loss": 0.8791,
+      "step": 8295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306847453521241e-06,
+      "loss": 0.8545,
+      "step": 8296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304590993546474e-06,
+      "loss": 0.88,
+      "step": 8297
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302334962705848e-06,
+      "loss": 0.8969,
+      "step": 8298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.300079361169347e-06,
+      "loss": 0.8598,
+      "step": 8299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297824189106926e-06,
+      "loss": 0.8074,
+      "step": 8300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295569446688506e-06,
+      "loss": 0.8584,
+      "step": 8301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.293315134083975e-06,
+      "loss": 0.8434,
+      "step": 8302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.291061251463191e-06,
+      "loss": 0.8817,
+      "step": 8303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288807798995978e-06,
+      "loss": 0.8777,
+      "step": 8304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.286554776852125e-06,
+      "loss": 0.9238,
+      "step": 8305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284302185201395e-06,
+      "loss": 0.8527,
+      "step": 8306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282050024213514e-06,
+      "loss": 0.9074,
+      "step": 8307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279798294058176e-06,
+      "loss": 0.8642,
+      "step": 8308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2775469949050455e-06,
+      "loss": 0.8801,
+      "step": 8309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.275296126923743e-06,
+      "loss": 0.8837,
+      "step": 8310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.273045690283878e-06,
+      "loss": 0.8387,
+      "step": 8311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.270795685155001e-06,
+      "loss": 0.8363,
+      "step": 8312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.268546111706661e-06,
+      "loss": 0.8217,
+      "step": 8313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266296970108339e-06,
+      "loss": 0.8531,
+      "step": 8314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.26404826052952e-06,
+      "loss": 0.846,
+      "step": 8315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.261799983139621e-06,
+      "loss": 0.8742,
+      "step": 8316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259552138108061e-06,
+      "loss": 0.9316,
+      "step": 8317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257304725604192e-06,
+      "loss": 0.8952,
+      "step": 8318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.255057745797368e-06,
+      "loss": 0.8593,
+      "step": 8319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.252811198856878e-06,
+      "loss": 0.821,
+      "step": 8320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.250565084952002e-06,
+      "loss": 0.8699,
+      "step": 8321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.248319404251976e-06,
+      "loss": 0.8931,
+      "step": 8322
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.246074156926002e-06,
+      "loss": 0.8772,
+      "step": 8323
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2438293431432665e-06,
+      "loss": 0.8912,
+      "step": 8324
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.241584963072895e-06,
+      "loss": 0.8926,
+      "step": 8325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.239341016884008e-06,
+      "loss": 0.8766,
+      "step": 8326
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.237097504745671e-06,
+      "loss": 0.8685,
+      "step": 8327
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2348544268269375e-06,
+      "loss": 0.8781,
+      "step": 8328
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.232611783296804e-06,
+      "loss": 0.8454,
+      "step": 8329
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.230369574324262e-06,
+      "loss": 0.8865,
+      "step": 8330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.228127800078246e-06,
+      "loss": 0.805,
+      "step": 8331
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225886460727671e-06,
+      "loss": 0.8488,
+      "step": 8332
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.223645556441416e-06,
+      "loss": 0.8519,
+      "step": 8333
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2214050873883285e-06,
+      "loss": 0.8362,
+      "step": 8334
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.21916505373722e-06,
+      "loss": 0.854,
+      "step": 8335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2169254556568714e-06,
+      "loss": 0.8434,
+      "step": 8336
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214686293316031e-06,
+      "loss": 0.8383,
+      "step": 8337
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.212447566883415e-06,
+      "loss": 0.8843,
+      "step": 8338
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210209276527705e-06,
+      "loss": 0.8805,
+      "step": 8339
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.207971422417549e-06,
+      "loss": 0.8584,
+      "step": 8340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.205734004721565e-06,
+      "loss": 0.8519,
+      "step": 8341
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203497023608335e-06,
+      "loss": 0.8564,
+      "step": 8342
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.201260479246411e-06,
+      "loss": 0.8457,
+      "step": 8343
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19902437180431e-06,
+      "loss": 0.8669,
+      "step": 8344
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196788701450517e-06,
+      "loss": 0.8749,
+      "step": 8345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194553468353485e-06,
+      "loss": 0.8705,
+      "step": 8346
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192318672681631e-06,
+      "loss": 0.9012,
+      "step": 8347
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190084314603342e-06,
+      "loss": 0.8398,
+      "step": 8348
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187850394286972e-06,
+      "loss": 0.8916,
+      "step": 8349
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1856169119008384e-06,
+      "loss": 0.9101,
+      "step": 8350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1833838676132346e-06,
+      "loss": 0.8462,
+      "step": 8351
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181151261592403e-06,
+      "loss": 0.8457,
+      "step": 8352
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.178919094006578e-06,
+      "loss": 0.8876,
+      "step": 8353
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176687365023934e-06,
+      "loss": 0.8814,
+      "step": 8354
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.174456074812641e-06,
+      "loss": 0.8412,
+      "step": 8355
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1722252235408045e-06,
+      "loss": 0.828,
+      "step": 8356
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.16999481137653e-06,
+      "loss": 0.8787,
+      "step": 8357
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167764838487857e-06,
+      "loss": 0.8519,
+      "step": 8358
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165535305042822e-06,
+      "loss": 0.8609,
+      "step": 8359
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163306211209404e-06,
+      "loss": 0.5007,
+      "step": 8360
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.161077557155569e-06,
+      "loss": 0.8184,
+      "step": 8361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.158849343049233e-06,
+      "loss": 0.8874,
+      "step": 8362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156621569058288e-06,
+      "loss": 0.8659,
+      "step": 8363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.15439423535059e-06,
+      "loss": 0.8452,
+      "step": 8364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.152167342093965e-06,
+      "loss": 0.7983,
+      "step": 8365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149940889456202e-06,
+      "loss": 0.8455,
+      "step": 8366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147714877605055e-06,
+      "loss": 0.5562,
+      "step": 8367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1454893067082605e-06,
+      "loss": 0.8719,
+      "step": 8368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.143264176933492e-06,
+      "loss": 0.85,
+      "step": 8369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141039488448424e-06,
+      "loss": 0.8306,
+      "step": 8370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.138815241420666e-06,
+      "loss": 0.8785,
+      "step": 8371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1365914360178235e-06,
+      "loss": 0.8619,
+      "step": 8372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.134368072407444e-06,
+      "loss": 0.8616,
+      "step": 8373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1321451507570555e-06,
+      "loss": 0.9041,
+      "step": 8374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129922671234148e-06,
+      "loss": 0.8806,
+      "step": 8375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.127700634006181e-06,
+      "loss": 0.8982,
+      "step": 8376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.12547903924058e-06,
+      "loss": 0.8607,
+      "step": 8377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.123257887104733e-06,
+      "loss": 0.8637,
+      "step": 8378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.121037177766002e-06,
+      "loss": 0.8663,
+      "step": 8379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118816911391709e-06,
+      "loss": 0.8492,
+      "step": 8380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116597088149146e-06,
+      "loss": 0.8362,
+      "step": 8381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1143777082055715e-06,
+      "loss": 0.7956,
+      "step": 8382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1121587717282085e-06,
+      "loss": 0.863,
+      "step": 8383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109940278884249e-06,
+      "loss": 0.8776,
+      "step": 8384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.10772222984085e-06,
+      "loss": 0.914,
+      "step": 8385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105504624765137e-06,
+      "loss": 0.8396,
+      "step": 8386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1032874638242004e-06,
+      "loss": 0.8966,
+      "step": 8387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1010707471850965e-06,
+      "loss": 0.9399,
+      "step": 8388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.098854475014849e-06,
+      "loss": 0.8899,
+      "step": 8389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09663864748045e-06,
+      "loss": 0.8733,
+      "step": 8390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094423264748855e-06,
+      "loss": 0.8612,
+      "step": 8391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.092208326986986e-06,
+      "loss": 0.8119,
+      "step": 8392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089993834361738e-06,
+      "loss": 0.851,
+      "step": 8393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087779787039957e-06,
+      "loss": 0.8979,
+      "step": 8394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085566185188478e-06,
+      "loss": 0.8689,
+      "step": 8395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083353028974079e-06,
+      "loss": 0.8762,
+      "step": 8396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.081140318563526e-06,
+      "loss": 0.9047,
+      "step": 8397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078928054123529e-06,
+      "loss": 0.8989,
+      "step": 8398
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.076716235820789e-06,
+      "loss": 0.8828,
+      "step": 8399
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074504863821949e-06,
+      "loss": 0.8267,
+      "step": 8400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.072293938293641e-06,
+      "loss": 0.8285,
+      "step": 8401
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.070083459402442e-06,
+      "loss": 0.85,
+      "step": 8402
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067873427314917e-06,
+      "loss": 0.8578,
+      "step": 8403
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065663842197576e-06,
+      "loss": 0.8228,
+      "step": 8404
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.06345470421691e-06,
+      "loss": 0.8847,
+      "step": 8405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061246013539372e-06,
+      "loss": 0.5378,
+      "step": 8406
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.059037770331379e-06,
+      "loss": 0.8766,
+      "step": 8407
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056829974759318e-06,
+      "loss": 0.8403,
+      "step": 8408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054622626989539e-06,
+      "loss": 0.8984,
+      "step": 8409
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0524157271883635e-06,
+      "loss": 0.8573,
+      "step": 8410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0502092755220676e-06,
+      "loss": 0.821,
+      "step": 8411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048003272156916e-06,
+      "loss": 0.8615,
+      "step": 8412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.045797717259109e-06,
+      "loss": 0.9136,
+      "step": 8413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.043592610994843e-06,
+      "loss": 0.593,
+      "step": 8414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041387953530258e-06,
+      "loss": 0.8601,
+      "step": 8415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0391837450314715e-06,
+      "loss": 0.8564,
+      "step": 8416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036979985664566e-06,
+      "loss": 0.8687,
+      "step": 8417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034776675595589e-06,
+      "loss": 0.8766,
+      "step": 8418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032573814990553e-06,
+      "loss": 0.8369,
+      "step": 8419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030371404015438e-06,
+      "loss": 0.9099,
+      "step": 8420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02816944283619e-06,
+      "loss": 0.8752,
+      "step": 8421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.025967931618722e-06,
+      "loss": 0.9093,
+      "step": 8422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02376687052891e-06,
+      "loss": 0.8289,
+      "step": 8423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0215662597326e-06,
+      "loss": 0.8321,
+      "step": 8424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019366099395602e-06,
+      "loss": 0.891,
+      "step": 8425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017166389683692e-06,
+      "loss": 0.874,
+      "step": 8426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.014967130762612e-06,
+      "loss": 0.9111,
+      "step": 8427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.012768322798072e-06,
+      "loss": 0.8802,
+      "step": 8428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010569965955744e-06,
+      "loss": 0.8877,
+      "step": 8429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.00837206040127e-06,
+      "loss": 0.8492,
+      "step": 8430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.006174606300255e-06,
+      "loss": 0.8744,
+      "step": 8431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003977603818274e-06,
+      "loss": 0.8657,
+      "step": 8432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.001781053120863e-06,
+      "loss": 0.8395,
+      "step": 8433
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999584954373528e-06,
+      "loss": 0.8993,
+      "step": 8434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99738930774174e-06,
+      "loss": 0.8264,
+      "step": 8435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.995194113390929e-06,
+      "loss": 0.8481,
+      "step": 8436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.992999371486508e-06,
+      "loss": 0.8084,
+      "step": 8437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.990805082193831e-06,
+      "loss": 0.8776,
+      "step": 8438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9886112456782475e-06,
+      "loss": 0.8944,
+      "step": 8439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.986417862105043e-06,
+      "loss": 0.8932,
+      "step": 8440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.984224931639496e-06,
+      "loss": 0.8832,
+      "step": 8441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9820324544468255e-06,
+      "loss": 0.8599,
+      "step": 8442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.979840430692242e-06,
+      "loss": 0.8616,
+      "step": 8443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.977648860540896e-06,
+      "loss": 0.8676,
+      "step": 8444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.975457744157929e-06,
+      "loss": 0.8752,
+      "step": 8445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.973267081708425e-06,
+      "loss": 0.8733,
+      "step": 8446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.97107687335745e-06,
+      "loss": 0.882,
+      "step": 8447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.96888711927003e-06,
+      "loss": 0.8855,
+      "step": 8448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9666978196111575e-06,
+      "loss": 0.8371,
+      "step": 8449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964508974545788e-06,
+      "loss": 0.8745,
+      "step": 8450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9623205842388475e-06,
+      "loss": 0.8282,
+      "step": 8451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960132648855226e-06,
+      "loss": 0.8693,
+      "step": 8452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957945168559777e-06,
+      "loss": 0.8365,
+      "step": 8453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9557581435173235e-06,
+      "loss": 0.8334,
+      "step": 8454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.953571573892646e-06,
+      "loss": 0.8514,
+      "step": 8455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.951385459850508e-06,
+      "loss": 0.8749,
+      "step": 8456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949199801555619e-06,
+      "loss": 0.9061,
+      "step": 8457
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947014599172664e-06,
+      "loss": 0.8813,
+      "step": 8458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.944829852866293e-06,
+      "loss": 0.877,
+      "step": 8459
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9426455628011204e-06,
+      "loss": 0.8804,
+      "step": 8460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.940461729141728e-06,
+      "loss": 0.9079,
+      "step": 8461
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93827835205266e-06,
+      "loss": 0.8527,
+      "step": 8462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93609543169843e-06,
+      "loss": 0.8563,
+      "step": 8463
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.933912968243515e-06,
+      "loss": 0.9072,
+      "step": 8464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9317309618523565e-06,
+      "loss": 0.8716,
+      "step": 8465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.929549412689368e-06,
+      "loss": 0.8297,
+      "step": 8466
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9273683209189115e-06,
+      "loss": 0.8481,
+      "step": 8467
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925187686705339e-06,
+      "loss": 0.8628,
+      "step": 8468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.923007510212952e-06,
+      "loss": 0.882,
+      "step": 8469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920827791606018e-06,
+      "loss": 0.887,
+      "step": 8470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.918648531048776e-06,
+      "loss": 0.8697,
+      "step": 8471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916469728705428e-06,
+      "loss": 0.9016,
+      "step": 8472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.914291384740139e-06,
+      "loss": 0.8793,
+      "step": 8473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.912113499317042e-06,
+      "loss": 0.8713,
+      "step": 8474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9099360726002355e-06,
+      "loss": 0.8858,
+      "step": 8475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907759104753782e-06,
+      "loss": 0.8836,
+      "step": 8476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905582595941716e-06,
+      "loss": 0.8702,
+      "step": 8477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903406546328019e-06,
+      "loss": 0.8589,
+      "step": 8478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.901230956076665e-06,
+      "loss": 0.8712,
+      "step": 8479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.899055825351567e-06,
+      "loss": 0.868,
+      "step": 8480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.896881154316625e-06,
+      "loss": 0.8588,
+      "step": 8481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894706943135686e-06,
+      "loss": 0.8373,
+      "step": 8482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.892533191972582e-06,
+      "loss": 0.8328,
+      "step": 8483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890359900991086e-06,
+      "loss": 0.8408,
+      "step": 8484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888187070354964e-06,
+      "loss": 0.8518,
+      "step": 8485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886014700227919e-06,
+      "loss": 0.9422,
+      "step": 8486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.883842790773647e-06,
+      "loss": 0.9219,
+      "step": 8487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881671342155786e-06,
+      "loss": 0.8752,
+      "step": 8488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879500354537952e-06,
+      "loss": 0.9162,
+      "step": 8489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8773298280837245e-06,
+      "loss": 0.8858,
+      "step": 8490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875159762956644e-06,
+      "loss": 0.838,
+      "step": 8491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872990159320222e-06,
+      "loss": 0.9141,
+      "step": 8492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.870821017337931e-06,
+      "loss": 0.8629,
+      "step": 8493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.868652337173211e-06,
+      "loss": 0.8876,
+      "step": 8494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866484118989465e-06,
+      "loss": 0.869,
+      "step": 8495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864316362950065e-06,
+      "loss": 0.8624,
+      "step": 8496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862149069218343e-06,
+      "loss": 0.857,
+      "step": 8497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8599822379576005e-06,
+      "loss": 0.839,
+      "step": 8498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.857815869331102e-06,
+      "loss": 0.8844,
+      "step": 8499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855649963502078e-06,
+      "loss": 0.8767,
+      "step": 8500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.853484520633725e-06,
+      "loss": 0.5088,
+      "step": 8501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8513195408892e-06,
+      "loss": 0.8678,
+      "step": 8502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8491550244316326e-06,
+      "loss": 0.8813,
+      "step": 8503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.84699097142411e-06,
+      "loss": 0.918,
+      "step": 8504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8448273820296915e-06,
+      "loss": 0.8438,
+      "step": 8505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842664256411393e-06,
+      "loss": 0.83,
+      "step": 8506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840501594732205e-06,
+      "loss": 0.8295,
+      "step": 8507
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8383393971550805e-06,
+      "loss": 0.8439,
+      "step": 8508
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836177663842925e-06,
+      "loss": 0.8488,
+      "step": 8509
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.834016394958632e-06,
+      "loss": 0.8538,
+      "step": 8510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.831855590665035e-06,
+      "loss": 0.8506,
+      "step": 8511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.829695251124953e-06,
+      "loss": 0.8514,
+      "step": 8512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827535376501162e-06,
+      "loss": 0.8569,
+      "step": 8513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8253759669564e-06,
+      "loss": 0.8809,
+      "step": 8514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.823217022653376e-06,
+      "loss": 0.8477,
+      "step": 8515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821058543754758e-06,
+      "loss": 0.8654,
+      "step": 8516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.818900530423181e-06,
+      "loss": 0.8876,
+      "step": 8517
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816742982821249e-06,
+      "loss": 0.8545,
+      "step": 8518
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8145859011115283e-06,
+      "loss": 0.8517,
+      "step": 8519
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8124292854565405e-06,
+      "loss": 0.8358,
+      "step": 8520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810273136018793e-06,
+      "loss": 0.8585,
+      "step": 8521
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8081174529607346e-06,
+      "loss": 0.8456,
+      "step": 8522
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8059622364448023e-06,
+      "loss": 0.8304,
+      "step": 8523
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803807486633373e-06,
+      "loss": 0.8765,
+      "step": 8524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801653203688814e-06,
+      "loss": 0.8974,
+      "step": 8525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7994993877734345e-06,
+      "loss": 0.8654,
+      "step": 8526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797346039049529e-06,
+      "loss": 0.8476,
+      "step": 8527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7951931576793333e-06,
+      "loss": 0.8468,
+      "step": 8528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.793040743825077e-06,
+      "loss": 0.8319,
+      "step": 8529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7908887976489284e-06,
+      "loss": 0.8588,
+      "step": 8530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7887373193130327e-06,
+      "loss": 0.7925,
+      "step": 8531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7865863089794985e-06,
+      "loss": 0.9069,
+      "step": 8532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7844357668104005e-06,
+      "loss": 0.8212,
+      "step": 8533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7822856929677753e-06,
+      "loss": 0.5546,
+      "step": 8534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.780136087613624e-06,
+      "loss": 0.859,
+      "step": 8535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7779869509099166e-06,
+      "loss": 0.8042,
+      "step": 8536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7758382830185826e-06,
+      "loss": 0.7955,
+      "step": 8537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.773690084101519e-06,
+      "loss": 0.873,
+      "step": 8538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7715423543205875e-06,
+      "loss": 0.8614,
+      "step": 8539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7693950938376145e-06,
+      "loss": 0.8161,
+      "step": 8540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.767248302814388e-06,
+      "loss": 0.8215,
+      "step": 8541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.8527,
+      "step": 8542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7629561297941665e-06,
+      "loss": 0.8297,
+      "step": 8543
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7608107481205735e-06,
+      "loss": 0.854,
+      "step": 8544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7586658365535367e-06,
+      "loss": 0.8678,
+      "step": 8545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75652139525467e-06,
+      "loss": 0.8224,
+      "step": 8546
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75437742438555e-06,
+      "loss": 0.8703,
+      "step": 8547
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.752233924107721e-06,
+      "loss": 0.8409,
+      "step": 8548
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7500908945826885e-06,
+      "loss": 0.8927,
+      "step": 8549
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.747948335971928e-06,
+      "loss": 0.9006,
+      "step": 8550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.745806248436866e-06,
+      "loss": 0.8596,
+      "step": 8551
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.743664632138918e-06,
+      "loss": 0.8072,
+      "step": 8552
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7415234872394323e-06,
+      "loss": 0.8809,
+      "step": 8553
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7393828138997543e-06,
+      "loss": 0.8842,
+      "step": 8554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7372426122811658e-06,
+      "loss": 0.4513,
+      "step": 8555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.735102882544932e-06,
+      "loss": 0.9056,
+      "step": 8556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.732963624852275e-06,
+      "loss": 0.8932,
+      "step": 8557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.730824839364382e-06,
+      "loss": 0.8474,
+      "step": 8558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7286865262424056e-06,
+      "loss": 0.8403,
+      "step": 8559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.72654868564746e-06,
+      "loss": 0.8587,
+      "step": 8560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7244113177406305e-06,
+      "loss": 0.8765,
+      "step": 8561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.722274422682952e-06,
+      "loss": 0.8418,
+      "step": 8562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.720138000635447e-06,
+      "loss": 0.8874,
+      "step": 8563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7180020517590764e-06,
+      "loss": 0.9166,
+      "step": 8564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.715866576214793e-06,
+      "loss": 0.847,
+      "step": 8565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7137315741634825e-06,
+      "loss": 0.8871,
+      "step": 8566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7115970457660277e-06,
+      "loss": 0.8397,
+      "step": 8567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7094629911832458e-06,
+      "loss": 0.8857,
+      "step": 8568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7073294105759462e-06,
+      "loss": 0.8492,
+      "step": 8569
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.705196304104873e-06,
+      "loss": 0.8885,
+      "step": 8570
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7030636719307666e-06,
+      "loss": 0.8819,
+      "step": 8571
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7009315142143033e-06,
+      "loss": 0.9088,
+      "step": 8572
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6987998311161388e-06,
+      "loss": 0.8426,
+      "step": 8573
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6966686227968907e-06,
+      "loss": 0.8613,
+      "step": 8574
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6945378894171392e-06,
+      "loss": 0.8798,
+      "step": 8575
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.69240763113743e-06,
+      "loss": 0.9003,
+      "step": 8576
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.690277848118273e-06,
+      "loss": 0.8575,
+      "step": 8577
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.688148540520141e-06,
+      "loss": 0.8593,
+      "step": 8578
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.686019708503472e-06,
+      "loss": 0.8692,
+      "step": 8579
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.683891352228668e-06,
+      "loss": 0.8914,
+      "step": 8580
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6817634718560947e-06,
+      "loss": 0.8928,
+      "step": 8581
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6796360675460817e-06,
+      "loss": 0.5316,
+      "step": 8582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6775091394589247e-06,
+      "loss": 0.8722,
+      "step": 8583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6753826877548817e-06,
+      "loss": 0.8955,
+      "step": 8584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6732567125941755e-06,
+      "loss": 0.833,
+      "step": 8585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6711312141369913e-06,
+      "loss": 0.864,
+      "step": 8586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6690061925434817e-06,
+      "loss": 0.8882,
+      "step": 8587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6668816479737614e-06,
+      "loss": 0.8445,
+      "step": 8588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6647575805879077e-06,
+      "loss": 0.8471,
+      "step": 8589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.662633990545964e-06,
+      "loss": 0.8793,
+      "step": 8590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6605108780079378e-06,
+      "loss": 0.8975,
+      "step": 8591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6583882431338047e-06,
+      "loss": 0.9067,
+      "step": 8592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6562660860834866e-06,
+      "loss": 0.8638,
+      "step": 8593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6541444070168984e-06,
+      "loss": 0.8374,
+      "step": 8594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6520232060938886e-06,
+      "loss": 0.8177,
+      "step": 8595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6499024834742967e-06,
+      "loss": 0.8413,
+      "step": 8596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.647782239317903e-06,
+      "loss": 0.8635,
+      "step": 8597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6456624737844726e-06,
+      "loss": 0.9059,
+      "step": 8598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6435431870337123e-06,
+      "loss": 0.8247,
+      "step": 8599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6414243792253146e-06,
+      "loss": 0.9022,
+      "step": 8600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.639306050518924e-06,
+      "loss": 0.8257,
+      "step": 8601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.637188201074149e-06,
+      "loss": 0.7988,
+      "step": 8602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6350708310505678e-06,
+      "loss": 0.8685,
+      "step": 8603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6329539406077097e-06,
+      "loss": 0.9131,
+      "step": 8604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.630837529905089e-06,
+      "loss": 0.881,
+      "step": 8605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6287215991021596e-06,
+      "loss": 0.8341,
+      "step": 8606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6266061483583635e-06,
+      "loss": 0.8811,
+      "step": 8607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6244911778330826e-06,
+      "loss": 0.8345,
+      "step": 8608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.622376687685686e-06,
+      "loss": 0.8803,
+      "step": 8609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6202626780754823e-06,
+      "loss": 0.8772,
+      "step": 8610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6181491491617706e-06,
+      "loss": 0.8594,
+      "step": 8611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6160361011037846e-06,
+      "loss": 0.8803,
+      "step": 8612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6139235340607516e-06,
+      "loss": 0.8434,
+      "step": 8613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.611811448191839e-06,
+      "loss": 0.8766,
+      "step": 8614
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6096998436561883e-06,
+      "loss": 0.8672,
+      "step": 8615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6075887206129047e-06,
+      "loss": 0.5599,
+      "step": 8616
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6054780792210542e-06,
+      "loss": 0.8666,
+      "step": 8617
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.60336791963967e-06,
+      "loss": 0.8791,
+      "step": 8618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6012582420277463e-06,
+      "loss": 0.8965,
+      "step": 8619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5991490465442413e-06,
+      "loss": 0.8862,
+      "step": 8620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5970403333480786e-06,
+      "loss": 0.8424,
+      "step": 8621
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.594932102598142e-06,
+      "loss": 0.8459,
+      "step": 8622
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5928243544532835e-06,
+      "loss": 0.8507,
+      "step": 8623
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5907170890723165e-06,
+      "loss": 0.8757,
+      "step": 8624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5886103066140156e-06,
+      "loss": 0.8884,
+      "step": 8625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5865040072371228e-06,
+      "loss": 0.8323,
+      "step": 8626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.584398191100341e-06,
+      "loss": 0.8723,
+      "step": 8627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5822928583623405e-06,
+      "loss": 0.908,
+      "step": 8628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.580188009181751e-06,
+      "loss": 0.871,
+      "step": 8629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5780836437171673e-06,
+      "loss": 0.8414,
+      "step": 8630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5759797621271475e-06,
+      "loss": 0.8562,
+      "step": 8631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5738763645702145e-06,
+      "loss": 0.884,
+      "step": 8632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.571773451204853e-06,
+      "loss": 0.838,
+      "step": 8633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.569671022189516e-06,
+      "loss": 0.893,
+      "step": 8634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5675690776826055e-06,
+      "loss": 0.8408,
+      "step": 8635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.565467617842511e-06,
+      "loss": 0.8729,
+      "step": 8636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5633666428275583e-06,
+      "loss": 0.864,
+      "step": 8637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5612661527960646e-06,
+      "loss": 0.5163,
+      "step": 8638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5591661479062823e-06,
+      "loss": 0.9005,
+      "step": 8639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5570666283164548e-06,
+      "loss": 0.9185,
+      "step": 8640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.554967594184762e-06,
+      "loss": 0.8716,
+      "step": 8641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5528690456693747e-06,
+      "loss": 0.8571,
+      "step": 8642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5507709829283976e-06,
+      "loss": 0.9033,
+      "step": 8643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5486734061199266e-06,
+      "loss": 0.8485,
+      "step": 8644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546576315402007e-06,
+      "loss": 0.4693,
+      "step": 8645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.54447971093264e-06,
+      "loss": 0.8266,
+      "step": 8646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5423835928698126e-06,
+      "loss": 0.854,
+      "step": 8647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540287961371448e-06,
+      "loss": 0.8906,
+      "step": 8648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5381928165954604e-06,
+      "loss": 0.8873,
+      "step": 8649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.536098158699699e-06,
+      "loss": 0.8716,
+      "step": 8650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.534003987842005e-06,
+      "loss": 0.8611,
+      "step": 8651
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.531910304180156e-06,
+      "loss": 0.4711,
+      "step": 8652
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.529817107871918e-06,
+      "loss": 0.8454,
+      "step": 8653
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5277243990749944e-06,
+      "loss": 0.8022,
+      "step": 8654
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.525632177947078e-06,
+      "loss": 0.8348,
+      "step": 8655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.523540444645804e-06,
+      "loss": 0.8644,
+      "step": 8656
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5214491993287815e-06,
+      "loss": 0.8372,
+      "step": 8657
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5193584421535797e-06,
+      "loss": 0.8534,
+      "step": 8658
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5172681732777335e-06,
+      "loss": 0.4916,
+      "step": 8659
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.515178392858737e-06,
+      "loss": 0.8555,
+      "step": 8660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5130891010540513e-06,
+      "loss": 0.9025,
+      "step": 8661
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.511000298021098e-06,
+      "loss": 0.8354,
+      "step": 8662
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5089119839172625e-06,
+      "loss": 0.8653,
+      "step": 8663
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5068241588998965e-06,
+      "loss": 0.8351,
+      "step": 8664
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.504736823126309e-06,
+      "loss": 0.8701,
+      "step": 8665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502649976753777e-06,
+      "loss": 0.8339,
+      "step": 8666
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5005636199395377e-06,
+      "loss": 0.8434,
+      "step": 8667
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4984777528407944e-06,
+      "loss": 0.8338,
+      "step": 8668
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.49639237561471e-06,
+      "loss": 0.8881,
+      "step": 8669
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4943074884184125e-06,
+      "loss": 0.9,
+      "step": 8670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.492223091408994e-06,
+      "loss": 0.8533,
+      "step": 8671
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4901391847435074e-06,
+      "loss": 0.8771,
+      "step": 8672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4880557685789696e-06,
+      "loss": 0.8843,
+      "step": 8673
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4859728430723595e-06,
+      "loss": 0.8347,
+      "step": 8674
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4838904083806225e-06,
+      "loss": 0.9064,
+      "step": 8675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4818084646606663e-06,
+      "loss": 0.872,
+      "step": 8676
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.479727012069349e-06,
+      "loss": 0.8709,
+      "step": 8677
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.477646050763519e-06,
+      "loss": 0.8848,
+      "step": 8678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4755655808999555e-06,
+      "loss": 0.8561,
+      "step": 8679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.47348560263543e-06,
+      "loss": 0.8262,
+      "step": 8680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4714061161266497e-06,
+      "loss": 0.8736,
+      "step": 8681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4693271215303138e-06,
+      "loss": 0.8793,
+      "step": 8682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4672486190030543e-06,
+      "loss": 0.8344,
+      "step": 8683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4651706087014936e-06,
+      "loss": 0.9119,
+      "step": 8684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4630930907821926e-06,
+      "loss": 0.8413,
+      "step": 8685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4610160654016987e-06,
+      "loss": 0.8979,
+      "step": 8686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.458939532716502e-06,
+      "loss": 0.8349,
+      "step": 8687
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4568634928830624e-06,
+      "loss": 0.884,
+      "step": 8688
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.454787946057814e-06,
+      "loss": 0.8354,
+      "step": 8689
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.452712892397132e-06,
+      "loss": 0.8515,
+      "step": 8690
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4506383320573778e-06,
+      "loss": 0.8881,
+      "step": 8691
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4485642651948516e-06,
+      "loss": 0.8968,
+      "step": 8692
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4464906919658413e-06,
+      "loss": 0.8572,
+      "step": 8693
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4444176125265736e-06,
+      "loss": 0.863,
+      "step": 8694
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4423450270332626e-06,
+      "loss": 0.8852,
+      "step": 8695
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4402729356420574e-06,
+      "loss": 0.8469,
+      "step": 8696
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4382013385090985e-06,
+      "loss": 0.8655,
+      "step": 8697
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4361302357904657e-06,
+      "loss": 0.8613,
+      "step": 8698
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4340596276422154e-06,
+      "loss": 0.8668,
+      "step": 8699
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4319895142203618e-06,
+      "loss": 0.8142,
+      "step": 8700
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.429919895680881e-06,
+      "loss": 0.8889,
+      "step": 8701
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.427850772179715e-06,
+      "loss": 0.859,
+      "step": 8702
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4257821438727667e-06,
+      "loss": 0.8255,
+      "step": 8703
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4237140109159015e-06,
+      "loss": 0.8583,
+      "step": 8704
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4216463734649486e-06,
+      "loss": 0.8433,
+      "step": 8705
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.419579231675698e-06,
+      "loss": 0.8512,
+      "step": 8706
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4175125857039027e-06,
+      "loss": 0.8247,
+      "step": 8707
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.415446435705282e-06,
+      "loss": 0.5402,
+      "step": 8708
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4133807818355115e-06,
+      "loss": 0.8751,
+      "step": 8709
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4113156242502345e-06,
+      "loss": 0.8554,
+      "step": 8710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409250963105055e-06,
+      "loss": 0.8456,
+      "step": 8711
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.407186798555541e-06,
+      "loss": 0.8887,
+      "step": 8712
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4051231307572187e-06,
+      "loss": 0.8694,
+      "step": 8713
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.403059959865583e-06,
+      "loss": 0.81,
+      "step": 8714
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4009972860360872e-06,
+      "loss": 0.9032,
+      "step": 8715
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3989351094241496e-06,
+      "loss": 0.8446,
+      "step": 8716
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.396873430185148e-06,
+      "loss": 0.8404,
+      "step": 8717
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3948122484744296e-06,
+      "loss": 0.8481,
+      "step": 8718
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3927515644472876e-06,
+      "loss": 0.8319,
+      "step": 8719
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.390691378259002e-06,
+      "loss": 0.8381,
+      "step": 8720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.388631690064792e-06,
+      "loss": 0.8464,
+      "step": 8721
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.38657250001986e-06,
+      "loss": 0.8554,
+      "step": 8722
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.384513808279348e-06,
+      "loss": 0.8874,
+      "step": 8723
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.382455614998388e-06,
+      "loss": 0.9131,
+      "step": 8724
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.380397920332045e-06,
+      "loss": 0.9237,
+      "step": 8725
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3783407244353726e-06,
+      "loss": 0.5515,
+      "step": 8726
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3762840274633647e-06,
+      "loss": 0.9091,
+      "step": 8727
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3742278295709996e-06,
+      "loss": 0.8179,
+      "step": 8728
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.372172130913196e-06,
+      "loss": 0.8446,
+      "step": 8729
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3701169316448502e-06,
+      "loss": 0.8734,
+      "step": 8730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3680622319208158e-06,
+      "loss": 0.8991,
+      "step": 8731
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3660080318959043e-06,
+      "loss": 0.8779,
+      "step": 8732
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3639543317249056e-06,
+      "loss": 0.8477,
+      "step": 8733
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.361901131562547e-06,
+      "loss": 0.8668,
+      "step": 8734
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.359848431563546e-06,
+      "loss": 0.8085,
+      "step": 8735
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.357796231882553e-06,
+      "loss": 0.8382,
+      "step": 8736
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355744532674211e-06,
+      "loss": 0.8906,
+      "step": 8737
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3536933340930956e-06,
+      "loss": 0.8476,
+      "step": 8738
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3516426362937717e-06,
+      "loss": 0.864,
+      "step": 8739
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3495924394307466e-06,
+      "loss": 0.8641,
+      "step": 8740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3475427436584985e-06,
+      "loss": 0.8005,
+      "step": 8741
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3454935491314678e-06,
+      "loss": 0.8687,
+      "step": 8742
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3434448560040544e-06,
+      "loss": 0.8364,
+      "step": 8743
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.341396664430624e-06,
+      "loss": 0.8669,
+      "step": 8744
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3393489745655016e-06,
+      "loss": 0.856,
+      "step": 8745
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3373017865629742e-06,
+      "loss": 0.8866,
+      "step": 8746
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.335255100577294e-06,
+      "loss": 0.9019,
+      "step": 8747
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3332089167626713e-06,
+      "loss": 0.8803,
+      "step": 8748
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3311632352732826e-06,
+      "loss": 0.8442,
+      "step": 8749
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.329118056263264e-06,
+      "loss": 0.8548,
+      "step": 8750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.327073379886714e-06,
+      "loss": 0.898,
+      "step": 8751
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.325029206297694e-06,
+      "loss": 0.8789,
+      "step": 8752
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.322985535650226e-06,
+      "loss": 0.8988,
+      "step": 8753
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3209423680982967e-06,
+      "loss": 0.5166,
+      "step": 8754
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3188997037958535e-06,
+      "loss": 0.928,
+      "step": 8755
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.316857542896804e-06,
+      "loss": 0.9251,
+      "step": 8756
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3148158855550217e-06,
+      "loss": 0.8913,
+      "step": 8757
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3127747319243385e-06,
+      "loss": 0.8143,
+      "step": 8758
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.310734082158552e-06,
+      "loss": 0.8726,
+      "step": 8759
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 0.8526,
+      "step": 8760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3066542948366564e-06,
+      "loss": 0.8734,
+      "step": 8761
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3046151575879526e-06,
+      "loss": 0.8518,
+      "step": 8762
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.30257652481894e-06,
+      "loss": 0.8599,
+      "step": 8763
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3005383966832383e-06,
+      "loss": 0.8789,
+      "step": 8764
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2985007733343997e-06,
+      "loss": 0.8556,
+      "step": 8765
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.29646365492597e-06,
+      "loss": 0.8643,
+      "step": 8766
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2944270416114256e-06,
+      "loss": 0.8403,
+      "step": 8767
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.292390933544234e-06,
+      "loss": 0.851,
+      "step": 8768
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.290355330877797e-06,
+      "loss": 0.8752,
+      "step": 8769
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288320233765504e-06,
+      "loss": 0.8634,
+      "step": 8770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.286285642360687e-06,
+      "loss": 0.8443,
+      "step": 8771
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.284251556816648e-06,
+      "loss": 0.8358,
+      "step": 8772
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.282217977286651e-06,
+      "loss": 0.8505,
+      "step": 8773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.280184903923921e-06,
+      "loss": 0.8547,
+      "step": 8774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.278152336881645e-06,
+      "loss": 0.8831,
+      "step": 8775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2761202763129674e-06,
+      "loss": 0.828,
+      "step": 8776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2740887223710094e-06,
+      "loss": 0.8505,
+      "step": 8777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2720576752088295e-06,
+      "loss": 0.8998,
+      "step": 8778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.270027134979474e-06,
+      "loss": 0.8936,
+      "step": 8779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.267997101835928e-06,
+      "loss": 0.9038,
+      "step": 8780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2659675759311615e-06,
+      "loss": 0.8477,
+      "step": 8781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2639385574180825e-06,
+      "loss": 0.837,
+      "step": 8782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.261910046449578e-06,
+      "loss": 0.8539,
+      "step": 8783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.259882043178488e-06,
+      "loss": 0.8493,
+      "step": 8784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2578545477576195e-06,
+      "loss": 0.8508,
+      "step": 8785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2558275603397392e-06,
+      "loss": 0.8638,
+      "step": 8786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2538010810775743e-06,
+      "loss": 0.8726,
+      "step": 8787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.251775110123814e-06,
+      "loss": 0.8273,
+      "step": 8788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2497496476311117e-06,
+      "loss": 0.8468,
+      "step": 8789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.247724693752079e-06,
+      "loss": 0.9164,
+      "step": 8790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.245700248639292e-06,
+      "loss": 0.8488,
+      "step": 8791
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2436763124452865e-06,
+      "loss": 0.8612,
+      "step": 8792
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2416528853225615e-06,
+      "loss": 0.8042,
+      "step": 8793
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2396299674235777e-06,
+      "loss": 0.8603,
+      "step": 8794
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.237607558900755e-06,
+      "loss": 0.8973,
+      "step": 8795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.235585659906476e-06,
+      "loss": 0.8551,
+      "step": 8796
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2335642705930883e-06,
+      "loss": 0.8549,
+      "step": 8797
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.231543391112896e-06,
+      "loss": 0.8737,
+      "step": 8798
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2295230216181673e-06,
+      "loss": 0.8403,
+      "step": 8799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.227503162261133e-06,
+      "loss": 0.8412,
+      "step": 8800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2254838131939836e-06,
+      "loss": 0.8696,
+      "step": 8801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.223464974568874e-06,
+      "loss": 0.8936,
+      "step": 8802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.22144664653791e-06,
+      "loss": 0.8429,
+      "step": 8803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2194288292531793e-06,
+      "loss": 0.8643,
+      "step": 8804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.217411522866707e-06,
+      "loss": 0.8381,
+      "step": 8805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.215394727530504e-06,
+      "loss": 0.8397,
+      "step": 8806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.213378443396519e-06,
+      "loss": 0.8656,
+      "step": 8807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2113626706166855e-06,
+      "loss": 0.8274,
+      "step": 8808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2093474093428733e-06,
+      "loss": 0.8774,
+      "step": 8809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2073326597269416e-06,
+      "loss": 0.8661,
+      "step": 8810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2053184219206823e-06,
+      "loss": 0.874,
+      "step": 8811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2033046960758763e-06,
+      "loss": 0.4773,
+      "step": 8812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.201291482344242e-06,
+      "loss": 0.8595,
+      "step": 8813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199278780877475e-06,
+      "loss": 0.886,
+      "step": 8814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.197266591827225e-06,
+      "loss": 0.8642,
+      "step": 8815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1952549153451064e-06,
+      "loss": 0.9076,
+      "step": 8816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.193243751582693e-06,
+      "loss": 0.8663,
+      "step": 8817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.191233100691521e-06,
+      "loss": 0.8722,
+      "step": 8818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1892229628230886e-06,
+      "loss": 0.8401,
+      "step": 8819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1872133381288494e-06,
+      "loss": 0.9037,
+      "step": 8820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1852042267602344e-06,
+      "loss": 0.7679,
+      "step": 8821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1831956288686116e-06,
+      "loss": 0.8856,
+      "step": 8822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1811875446053374e-06,
+      "loss": 0.8361,
+      "step": 8823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1791799741217046e-06,
+      "loss": 0.9221,
+      "step": 8824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1771729175689814e-06,
+      "loss": 0.8056,
+      "step": 8825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1751663750983954e-06,
+      "loss": 0.8189,
+      "step": 8826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.173160346861134e-06,
+      "loss": 0.8974,
+      "step": 8827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171154833008345e-06,
+      "loss": 0.8575,
+      "step": 8828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.169149833691141e-06,
+      "loss": 0.8279,
+      "step": 8829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1671453490605906e-06,
+      "loss": 0.8352,
+      "step": 8830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1651413792677287e-06,
+      "loss": 0.8873,
+      "step": 8831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1631379244635487e-06,
+      "loss": 0.9068,
+      "step": 8832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1611349847990083e-06,
+      "loss": 0.8953,
+      "step": 8833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1591325604250144e-06,
+      "loss": 0.8406,
+      "step": 8834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1571306514924537e-06,
+      "loss": 0.8438,
+      "step": 8835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1551292581521632e-06,
+      "loss": 0.8225,
+      "step": 8836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.153128380554941e-06,
+      "loss": 0.855,
+      "step": 8837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1511280188515468e-06,
+      "loss": 0.8929,
+      "step": 8838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.149128173192706e-06,
+      "loss": 0.9052,
+      "step": 8839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1471288437290993e-06,
+      "loss": 0.8633,
+      "step": 8840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1451300306113706e-06,
+      "loss": 0.8652,
+      "step": 8841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1431317339901267e-06,
+      "loss": 0.8583,
+      "step": 8842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141133954015935e-06,
+      "loss": 0.8717,
+      "step": 8843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.139136690839323e-06,
+      "loss": 0.8519,
+      "step": 8844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.137139944610772e-06,
+      "loss": 0.8674,
+      "step": 8845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.135143715480743e-06,
+      "loss": 0.8993,
+      "step": 8846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1331480035996352e-06,
+      "loss": 0.8586,
+      "step": 8847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1311528091178324e-06,
+      "loss": 0.8486,
+      "step": 8848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1291581321856533e-06,
+      "loss": 0.8896,
+      "step": 8849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1271639729534076e-06,
+      "loss": 0.866,
+      "step": 8850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1251703315713333e-06,
+      "loss": 0.8415,
+      "step": 8851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1231772081896616e-06,
+      "loss": 0.8944,
+      "step": 8852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1211846029585555e-06,
+      "loss": 0.8666,
+      "step": 8853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1191925160281644e-06,
+      "loss": 0.883,
+      "step": 8854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.117200947548579e-06,
+      "loss": 0.8718,
+      "step": 8855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.115209897669861e-06,
+      "loss": 0.848,
+      "step": 8856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1132193665420306e-06,
+      "loss": 0.8483,
+      "step": 8857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1112293543150695e-06,
+      "loss": 0.8763,
+      "step": 8858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1092398611389196e-06,
+      "loss": 0.8759,
+      "step": 8859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1072508871634843e-06,
+      "loss": 0.8865,
+      "step": 8860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1052624325386272e-06,
+      "loss": 0.8784,
+      "step": 8861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.103274497414174e-06,
+      "loss": 0.8537,
+      "step": 8862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1012870819399087e-06,
+      "loss": 0.8681,
+      "step": 8863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0993001862655768e-06,
+      "loss": 0.846,
+      "step": 8864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.097313810540894e-06,
+      "loss": 0.9334,
+      "step": 8865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.095327954915519e-06,
+      "loss": 0.8381,
+      "step": 8866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0933426195390838e-06,
+      "loss": 0.8758,
+      "step": 8867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091357804561178e-06,
+      "loss": 0.8966,
+      "step": 8868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.089373510131354e-06,
+      "loss": 0.8589,
+      "step": 8869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0873897363991214e-06,
+      "loss": 0.8787,
+      "step": 8870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.085406483513953e-06,
+      "loss": 0.848,
+      "step": 8871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0834237516252817e-06,
+      "loss": 0.8938,
+      "step": 8872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0814415408825005e-06,
+      "loss": 0.9239,
+      "step": 8873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.079459851434965e-06,
+      "loss": 0.8225,
+      "step": 8874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0774786834319923e-06,
+      "loss": 0.8336,
+      "step": 8875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0754980370228504e-06,
+      "loss": 0.8394,
+      "step": 8876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.073517912356786e-06,
+      "loss": 0.8473,
+      "step": 8877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0715383095829853e-06,
+      "loss": 0.8187,
+      "step": 8878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0695592288506148e-06,
+      "loss": 0.8705,
+      "step": 8879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.067580670308792e-06,
+      "loss": 0.905,
+      "step": 8880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0656026341065927e-06,
+      "loss": 0.8924,
+      "step": 8881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.063625120393059e-06,
+      "loss": 0.9247,
+      "step": 8882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.061648129317192e-06,
+      "loss": 0.8085,
+      "step": 8883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.059671661027951e-06,
+      "loss": 0.9063,
+      "step": 8884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0576957156742583e-06,
+      "loss": 0.8542,
+      "step": 8885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055720293404999e-06,
+      "loss": 0.4711,
+      "step": 8886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0537453943690076e-06,
+      "loss": 0.5173,
+      "step": 8887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0517710187150994e-06,
+      "loss": 0.8524,
+      "step": 8888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0497971665920257e-06,
+      "loss": 0.8397,
+      "step": 8889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.047823838148525e-06,
+      "loss": 0.8469,
+      "step": 8890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.045851033533268e-06,
+      "loss": 0.8428,
+      "step": 8891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0438787528949142e-06,
+      "loss": 0.7733,
+      "step": 8892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.041906996382056e-06,
+      "loss": 0.823,
+      "step": 8893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0399357641432735e-06,
+      "loss": 0.8295,
+      "step": 8894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0379650563270825e-06,
+      "loss": 0.8579,
+      "step": 8895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.035994873081981e-06,
+      "loss": 0.8361,
+      "step": 8896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0340252145564086e-06,
+      "loss": 0.8632,
+      "step": 8897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0320560808987774e-06,
+      "loss": 0.8785,
+      "step": 8898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.030087472257456e-06,
+      "loss": 0.8852,
+      "step": 8899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0281193887807737e-06,
+      "loss": 0.8433,
+      "step": 8900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.026151830617021e-06,
+      "loss": 0.8625,
+      "step": 8901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024184797914449e-06,
+      "loss": 0.816,
+      "step": 8902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.022218290821267e-06,
+      "loss": 0.8898,
+      "step": 8903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.020252309485646e-06,
+      "loss": 0.8785,
+      "step": 8904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.01828685405572e-06,
+      "loss": 0.8817,
+      "step": 8905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0163219246795772e-06,
+      "loss": 0.8436,
+      "step": 8906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0143575215052732e-06,
+      "loss": 0.8201,
+      "step": 8907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.012393644680819e-06,
+      "loss": 0.8595,
+      "step": 8908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0104302943541872e-06,
+      "loss": 0.8761,
+      "step": 8909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.008467470673312e-06,
+      "loss": 0.8225,
+      "step": 8910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0065051737860863e-06,
+      "loss": 0.901,
+      "step": 8911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0045434038403653e-06,
+      "loss": 0.8998,
+      "step": 8912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.002582160983962e-06,
+      "loss": 0.82,
+      "step": 8913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.000621445364651e-06,
+      "loss": 0.878,
+      "step": 8914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9986612571301675e-06,
+      "loss": 0.8594,
+      "step": 8915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9967015964282077e-06,
+      "loss": 0.8402,
+      "step": 8916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.994742463406427e-06,
+      "loss": 0.839,
+      "step": 8917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9927838582124334e-06,
+      "loss": 0.8985,
+      "step": 8918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9908257809938167e-06,
+      "loss": 0.9209,
+      "step": 8919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9888682318980975e-06,
+      "loss": 0.8368,
+      "step": 8920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9869112110727862e-06,
+      "loss": 0.888,
+      "step": 8921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9849547186653262e-06,
+      "loss": 0.8973,
+      "step": 8922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9829987548231432e-06,
+      "loss": 0.8631,
+      "step": 8923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9810433196936108e-06,
+      "loss": 0.8072,
+      "step": 8924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9790884134240672e-06,
+      "loss": 0.9137,
+      "step": 8925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9771340361618075e-06,
+      "loss": 0.8397,
+      "step": 8926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9751801880540907e-06,
+      "loss": 0.821,
+      "step": 8927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.973226869248136e-06,
+      "loss": 0.8939,
+      "step": 8928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.971274079891112e-06,
+      "loss": 0.8543,
+      "step": 8929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9693218201301687e-06,
+      "loss": 0.8243,
+      "step": 8930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9673700901123914e-06,
+      "loss": 0.8717,
+      "step": 8931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.96541888998485e-06,
+      "loss": 0.8774,
+      "step": 8932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9634682198945508e-06,
+      "loss": 0.9008,
+      "step": 8933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9615180799884823e-06,
+      "loss": 0.9088,
+      "step": 8934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.959568470413572e-06,
+      "loss": 0.8401,
+      "step": 8935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9576193913167285e-06,
+      "loss": 0.8987,
+      "step": 8936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9556708428447987e-06,
+      "loss": 0.847,
+      "step": 8937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9537228251446125e-06,
+      "loss": 0.7964,
+      "step": 8938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9517753383629388e-06,
+      "loss": 0.6093,
+      "step": 8939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9498283826465177e-06,
+      "loss": 0.776,
+      "step": 8940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9478819581420493e-06,
+      "loss": 0.8943,
+      "step": 8941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.94593606499619e-06,
+      "loss": 0.8449,
+      "step": 8942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.943990703355558e-06,
+      "loss": 0.8685,
+      "step": 8943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.942045873366731e-06,
+      "loss": 0.8268,
+      "step": 8944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9401015751762484e-06,
+      "loss": 0.5708,
+      "step": 8945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9381578089306062e-06,
+      "loss": 0.8779,
+      "step": 8946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9362145747762626e-06,
+      "loss": 0.8524,
+      "step": 8947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.934271872859635e-06,
+      "loss": 0.7897,
+      "step": 8948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9323297033271014e-06,
+      "loss": 0.8768,
+      "step": 8949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9303880663249985e-06,
+      "loss": 0.8215,
+      "step": 8950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.928446961999625e-06,
+      "loss": 0.9201,
+      "step": 8951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.926506390497236e-06,
+      "loss": 0.8906,
+      "step": 8952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.92456635196405e-06,
+      "loss": 0.8116,
+      "step": 8953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9226268465462426e-06,
+      "loss": 0.8762,
+      "step": 8954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9206878743899525e-06,
+      "loss": 0.856,
+      "step": 8955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.918749435641274e-06,
+      "loss": 0.8119,
+      "step": 8956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9168115304462643e-06,
+      "loss": 0.8316,
+      "step": 8957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.914874158950939e-06,
+      "loss": 0.8255,
+      "step": 8958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.912937321301278e-06,
+      "loss": 0.8633,
+      "step": 8959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9110010176432067e-06,
+      "loss": 0.8917,
+      "step": 8960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9090652481226334e-06,
+      "loss": 0.8633,
+      "step": 8961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9071300128854007e-06,
+      "loss": 0.8925,
+      "step": 8962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9051953120773367e-06,
+      "loss": 0.887,
+      "step": 8963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9032611458442028e-06,
+      "loss": 0.843,
+      "step": 8964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9013275143317453e-06,
+      "loss": 0.852,
+      "step": 8965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8993944176856458e-06,
+      "loss": 0.8807,
+      "step": 8966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.897461856051569e-06,
+      "loss": 0.8349,
+      "step": 8967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8955298295751245e-06,
+      "loss": 0.8638,
+      "step": 8968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.893598338401884e-06,
+      "loss": 0.8747,
+      "step": 8969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8916673826773856e-06,
+      "loss": 0.8383,
+      "step": 8970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8897369625471105e-06,
+      "loss": 0.8247,
+      "step": 8971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8878070781565247e-06,
+      "loss": 0.8668,
+      "step": 8972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8858777296510276e-06,
+      "loss": 0.8461,
+      "step": 8973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8839489171760015e-06,
+      "loss": 0.804,
+      "step": 8974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.882020640876766e-06,
+      "loss": 0.8298,
+      "step": 8975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8800929008986234e-06,
+      "loss": 0.853,
+      "step": 8976
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.878165697386812e-06,
+      "loss": 0.8758,
+      "step": 8977
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.876239030486554e-06,
+      "loss": 0.8524,
+      "step": 8978
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874312900343006e-06,
+      "loss": 0.8899,
+      "step": 8979
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.87238730710131e-06,
+      "loss": 0.8355,
+      "step": 8980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8704622509065437e-06,
+      "loss": 0.8528,
+      "step": 8981
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8685377319037586e-06,
+      "loss": 0.8194,
+      "step": 8982
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8666137502379632e-06,
+      "loss": 0.8106,
+      "step": 8983
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.864690306054123e-06,
+      "loss": 0.8555,
+      "step": 8984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.862767399497166e-06,
+      "loss": 0.8591,
+      "step": 8985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8608450307119772e-06,
+      "loss": 0.8877,
+      "step": 8986
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8589231998434032e-06,
+      "loss": 0.8465,
+      "step": 8987
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8570019070362464e-06,
+      "loss": 0.4616,
+      "step": 8988
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8550811524352727e-06,
+      "loss": 0.886,
+      "step": 8989
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8531609361852074e-06,
+      "loss": 0.8983,
+      "step": 8990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8512412584307314e-06,
+      "loss": 0.886,
+      "step": 8991
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8493221193164886e-06,
+      "loss": 0.8701,
+      "step": 8992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.847403518987081e-06,
+      "loss": 0.8771,
+      "step": 8993
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8454854575870694e-06,
+      "loss": 0.8707,
+      "step": 8994
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8435679352609747e-06,
+      "loss": 0.8345,
+      "step": 8995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.841650952153279e-06,
+      "loss": 0.8391,
+      "step": 8996
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8397345084084193e-06,
+      "loss": 0.8722,
+      "step": 8997
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8378186041707977e-06,
+      "loss": 0.8837,
+      "step": 8998
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8359032395847698e-06,
+      "loss": 0.8887,
+      "step": 8999
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.833988414794654e-06,
+      "loss": 0.5256,
+      "step": 9000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8320741299447306e-06,
+      "loss": 0.8461,
+      "step": 9001
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.830160385179228e-06,
+      "loss": 0.8012,
+      "step": 9002
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.828247180642353e-06,
+      "loss": 0.887,
+      "step": 9003
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8263345164782473e-06,
+      "loss": 0.8651,
+      "step": 9004
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8244223928310388e-06,
+      "loss": 0.8689,
+      "step": 9005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8225108098447885e-06,
+      "loss": 0.8563,
+      "step": 9006
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.82059976766354e-06,
+      "loss": 0.9014,
+      "step": 9007
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8186892664312758e-06,
+      "loss": 0.8814,
+      "step": 9008
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.816779306291957e-06,
+      "loss": 0.8567,
+      "step": 9009
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.814869887389483e-06,
+      "loss": 0.8644,
+      "step": 9010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.812961009867733e-06,
+      "loss": 0.8578,
+      "step": 9011
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8110526738705345e-06,
+      "loss": 0.858,
+      "step": 9012
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.809144879541669e-06,
+      "loss": 0.8418,
+      "step": 9013
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8072376270248924e-06,
+      "loss": 0.8813,
+      "step": 9014
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.805330916463902e-06,
+      "loss": 0.8657,
+      "step": 9015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8034247480023735e-06,
+      "loss": 0.8463,
+      "step": 9016
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.801519121783921e-06,
+      "loss": 0.8568,
+      "step": 9017
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7996140379521407e-06,
+      "loss": 0.8779,
+      "step": 9018
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7977094966505624e-06,
+      "loss": 0.8548,
+      "step": 9019
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.795805498022701e-06,
+      "loss": 0.8561,
+      "step": 9020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7939020422120065e-06,
+      "loss": 0.8873,
+      "step": 9021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.79199912936191e-06,
+      "loss": 0.8547,
+      "step": 9022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7900967596157823e-06,
+      "loss": 0.8707,
+      "step": 9023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7881949331169655e-06,
+      "loss": 0.8396,
+      "step": 9024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7862936500087566e-06,
+      "loss": 0.7937,
+      "step": 9025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7843929104344134e-06,
+      "loss": 0.8498,
+      "step": 9026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7824927145371507e-06,
+      "loss": 0.8734,
+      "step": 9027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7805930624601427e-06,
+      "loss": 0.8726,
+      "step": 9028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.778693954346525e-06,
+      "loss": 0.8266,
+      "step": 9029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7767953903393894e-06,
+      "loss": 0.8883,
+      "step": 9030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.774897370581787e-06,
+      "loss": 0.895,
+      "step": 9031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7729998952167304e-06,
+      "loss": 0.8711,
+      "step": 9032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.771102964387188e-06,
+      "loss": 0.813,
+      "step": 9033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7692065782360876e-06,
+      "loss": 0.588,
+      "step": 9034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7673107369063203e-06,
+      "loss": 0.8796,
+      "step": 9035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7654154405407297e-06,
+      "loss": 0.8763,
+      "step": 9036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.763520689282122e-06,
+      "loss": 0.8422,
+      "step": 9037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.761626483273262e-06,
+      "loss": 0.8619,
+      "step": 9038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7597328226568743e-06,
+      "loss": 0.8132,
+      "step": 9039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7578397075756404e-06,
+      "loss": 0.817,
+      "step": 9040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7559471381722004e-06,
+      "loss": 0.8735,
+      "step": 9041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.754055114589156e-06,
+      "loss": 0.8676,
+      "step": 9042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7521636369690687e-06,
+      "loss": 0.8609,
+      "step": 9043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7502727054544477e-06,
+      "loss": 0.8567,
+      "step": 9044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7483823201877813e-06,
+      "loss": 0.8245,
+      "step": 9045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7464924813114926e-06,
+      "loss": 0.8679,
+      "step": 9046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7446031889679893e-06,
+      "loss": 0.8791,
+      "step": 9047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.742714443299612e-06,
+      "loss": 0.8906,
+      "step": 9048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7408262444486844e-06,
+      "loss": 0.8466,
+      "step": 9049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7389385925574654e-06,
+      "loss": 0.8452,
+      "step": 9050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7370514877681965e-06,
+      "loss": 0.8285,
+      "step": 9051
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7351649302230553e-06,
+      "loss": 0.8716,
+      "step": 9052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7332789200641995e-06,
+      "loss": 0.8504,
+      "step": 9053
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7313934574337263e-06,
+      "loss": 0.8455,
+      "step": 9054
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.729508542473702e-06,
+      "loss": 0.9119,
+      "step": 9055
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.727624175326157e-06,
+      "loss": 0.8169,
+      "step": 9056
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.725740356133062e-06,
+      "loss": 0.8047,
+      "step": 9057
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.72385708503637e-06,
+      "loss": 0.8237,
+      "step": 9058
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7219743621779683e-06,
+      "loss": 0.8659,
+      "step": 9059
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7200921876997264e-06,
+      "loss": 0.8535,
+      "step": 9060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7182105617434516e-06,
+      "loss": 0.8592,
+      "step": 9061
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.716329484450929e-06,
+      "loss": 0.5627,
+      "step": 9062
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7144489559638833e-06,
+      "loss": 0.8515,
+      "step": 9063
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7125689764240173e-06,
+      "loss": 0.9024,
+      "step": 9064
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.710689545972973e-06,
+      "loss": 0.8135,
+      "step": 9065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.708810664752366e-06,
+      "loss": 0.7772,
+      "step": 9066
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7069323329037632e-06,
+      "loss": 0.8256,
+      "step": 9067
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.705054550568693e-06,
+      "loss": 0.8497,
+      "step": 9068
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7031773178886413e-06,
+      "loss": 0.8773,
+      "step": 9069
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701300635005052e-06,
+      "loss": 0.8642,
+      "step": 9070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.699424502059328e-06,
+      "loss": 0.8923,
+      "step": 9071
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.697548919192832e-06,
+      "loss": 0.8136,
+      "step": 9072
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6956738865468832e-06,
+      "loss": 0.8399,
+      "step": 9073
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6937994042627622e-06,
+      "loss": 0.8676,
+      "step": 9074
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6919254724817048e-06,
+      "loss": 0.8613,
+      "step": 9075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.690052091344907e-06,
+      "loss": 0.8717,
+      "step": 9076
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.688179260993523e-06,
+      "loss": 0.8567,
+      "step": 9077
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.686306981568666e-06,
+      "loss": 0.5431,
+      "step": 9078
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6844352532114084e-06,
+      "loss": 0.8335,
+      "step": 9079
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.682564076062778e-06,
+      "loss": 0.8601,
+      "step": 9080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6806934502637637e-06,
+      "loss": 0.8703,
+      "step": 9081
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.678823375955314e-06,
+      "loss": 0.8761,
+      "step": 9082
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6769538532783314e-06,
+      "loss": 0.8431,
+      "step": 9083
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6750848823736817e-06,
+      "loss": 0.874,
+      "step": 9084
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.673216463382189e-06,
+      "loss": 0.8471,
+      "step": 9085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.671348596444624e-06,
+      "loss": 0.8933,
+      "step": 9086
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.669481281701739e-06,
+      "loss": 0.8051,
+      "step": 9087
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6676145192942194e-06,
+      "loss": 0.8889,
+      "step": 9088
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6657483093627312e-06,
+      "loss": 0.8361,
+      "step": 9089
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6638826520478787e-06,
+      "loss": 0.8326,
+      "step": 9090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6620175474902444e-06,
+      "loss": 0.8674,
+      "step": 9091
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6601529958303475e-06,
+      "loss": 0.8852,
+      "step": 9092
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.658288997208689e-06,
+      "loss": 0.8802,
+      "step": 9093
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6564255517657044e-06,
+      "loss": 0.8953,
+      "step": 9094
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.654562659641812e-06,
+      "loss": 0.8168,
+      "step": 9095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6527003209773663e-06,
+      "loss": 0.8797,
+      "step": 9096
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.650838535912692e-06,
+      "loss": 0.8038,
+      "step": 9097
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.648977304588072e-06,
+      "loss": 0.8563,
+      "step": 9098
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6471166271437397e-06,
+      "loss": 0.4704,
+      "step": 9099
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.645256503719902e-06,
+      "loss": 0.8672,
+      "step": 9100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6433969344567033e-06,
+      "loss": 0.8571,
+      "step": 9101
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.641537919494268e-06,
+      "loss": 0.8425,
+      "step": 9102
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.639679458972657e-06,
+      "loss": 0.8523,
+      "step": 9103
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6378215530319117e-06,
+      "loss": 0.8571,
+      "step": 9104
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.635964201812008e-06,
+      "loss": 0.879,
+      "step": 9105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.634107405452906e-06,
+      "loss": 0.8321,
+      "step": 9106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6322511640945e-06,
+      "loss": 0.8706,
+      "step": 9107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6303954778766573e-06,
+      "loss": 0.8922,
+      "step": 9108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.628540346939198e-06,
+      "loss": 0.8509,
+      "step": 9109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6266857714219007e-06,
+      "loss": 0.8543,
+      "step": 9110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.624831751464505e-06,
+      "loss": 0.825,
+      "step": 9111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6229782872067042e-06,
+      "loss": 0.8737,
+      "step": 9112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.621125378788153e-06,
+      "loss": 0.841,
+      "step": 9113
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6192730263484633e-06,
+      "loss": 0.8727,
+      "step": 9114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.617421230027205e-06,
+      "loss": 0.8536,
+      "step": 9115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.615569989963905e-06,
+      "loss": 0.8819,
+      "step": 9116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6137193062980506e-06,
+      "loss": 0.8484,
+      "step": 9117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6118691791690865e-06,
+      "loss": 0.8992,
+      "step": 9118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6100196087164132e-06,
+      "loss": 0.4865,
+      "step": 9119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.608170595079391e-06,
+      "loss": 0.5255,
+      "step": 9120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60632213839734e-06,
+      "loss": 0.8703,
+      "step": 9121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6044742388095366e-06,
+      "loss": 0.8555,
+      "step": 9122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6026268964552125e-06,
+      "loss": 0.8787,
+      "step": 9123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.600780111473563e-06,
+      "loss": 0.8718,
+      "step": 9124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598933884003737e-06,
+      "loss": 0.8456,
+      "step": 9125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597088214184843e-06,
+      "loss": 0.8711,
+      "step": 9126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.595243102155951e-06,
+      "loss": 0.9097,
+      "step": 9127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.593398548056075e-06,
+      "loss": 0.8403,
+      "step": 9128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5915545520242114e-06,
+      "loss": 0.8934,
+      "step": 9129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.589711114199287e-06,
+      "loss": 0.8645,
+      "step": 9130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.587868234720212e-06,
+      "loss": 0.8791,
+      "step": 9131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5860259137258316e-06,
+      "loss": 0.8435,
+      "step": 9132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5841841513549703e-06,
+      "loss": 0.8452,
+      "step": 9133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.58234294774639e-06,
+      "loss": 0.849,
+      "step": 9134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.580502303038831e-06,
+      "loss": 0.8362,
+      "step": 9135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5786622173709695e-06,
+      "loss": 0.8476,
+      "step": 9136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576822690881463e-06,
+      "loss": 0.9021,
+      "step": 9137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.574983723708906e-06,
+      "loss": 0.8622,
+      "step": 9138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573145315991864e-06,
+      "loss": 0.8658,
+      "step": 9139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.571307467868854e-06,
+      "loss": 0.8952,
+      "step": 9140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.569470179478355e-06,
+      "loss": 0.9002,
+      "step": 9141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.567633450958801e-06,
+      "loss": 0.8731,
+      "step": 9142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.565797282448582e-06,
+      "loss": 0.8533,
+      "step": 9143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5639616740860573e-06,
+      "loss": 0.8397,
+      "step": 9144
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.562126626009522e-06,
+      "loss": 0.821,
+      "step": 9145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5602921383572577e-06,
+      "loss": 0.8191,
+      "step": 9146
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5584582112674717e-06,
+      "loss": 0.8442,
+      "step": 9147
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.55662484487836e-06,
+      "loss": 0.8679,
+      "step": 9148
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5547920393280547e-06,
+      "loss": 0.8913,
+      "step": 9149
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5529597947546526e-06,
+      "loss": 0.8741,
+      "step": 9150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5511281112962096e-06,
+      "loss": 0.7987,
+      "step": 9151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5492969890907383e-06,
+      "loss": 0.838,
+      "step": 9152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5474664282762098e-06,
+      "loss": 0.9152,
+      "step": 9153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.545636428990551e-06,
+      "loss": 0.9009,
+      "step": 9154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5438069913716477e-06,
+      "loss": 0.828,
+      "step": 9155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.541978115557344e-06,
+      "loss": 0.8384,
+      "step": 9156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.540149801685441e-06,
+      "loss": 0.8445,
+      "step": 9157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5383220498936957e-06,
+      "loss": 0.8478,
+      "step": 9158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.536494860319826e-06,
+      "loss": 0.8796,
+      "step": 9159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.534668233101505e-06,
+      "loss": 0.8318,
+      "step": 9160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532842168376365e-06,
+      "loss": 0.8883,
+      "step": 9161
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5310166662819947e-06,
+      "loss": 0.8233,
+      "step": 9162
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5291917269559408e-06,
+      "loss": 0.8653,
+      "step": 9163
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5273673505357078e-06,
+      "loss": 0.8517,
+      "step": 9164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5255435371587578e-06,
+      "loss": 0.8339,
+      "step": 9165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.52372028696251e-06,
+      "loss": 0.8603,
+      "step": 9166
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.521897600084342e-06,
+      "loss": 0.8648,
+      "step": 9167
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5200754766615876e-06,
+      "loss": 0.8454,
+      "step": 9168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5182539168315435e-06,
+      "loss": 0.8874,
+      "step": 9169
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5164329207314495e-06,
+      "loss": 0.848,
+      "step": 9170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5146124884985255e-06,
+      "loss": 0.8561,
+      "step": 9171
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.512792620269924e-06,
+      "loss": 0.8369,
+      "step": 9172
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5109733161827786e-06,
+      "loss": 0.9062,
+      "step": 9173
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5091545763741578e-06,
+      "loss": 0.8754,
+      "step": 9174
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5073364009811107e-06,
+      "loss": 0.8791,
+      "step": 9175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5055187901406208e-06,
+      "loss": 0.7989,
+      "step": 9176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5037017439896503e-06,
+      "loss": 0.8653,
+      "step": 9177
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.501885262665099e-06,
+      "loss": 0.8836,
+      "step": 9178
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.500069346303845e-06,
+      "loss": 0.8734,
+      "step": 9179
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4982539950427044e-06,
+      "loss": 0.8338,
+      "step": 9180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496439209018461e-06,
+      "loss": 0.8423,
+      "step": 9181
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.494624988367855e-06,
+      "loss": 0.8475,
+      "step": 9182
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.492811333227583e-06,
+      "loss": 0.8742,
+      "step": 9183
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4909982437342993e-06,
+      "loss": 0.9156,
+      "step": 9184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.489185720024616e-06,
+      "loss": 0.8269,
+      "step": 9185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.487373762235101e-06,
+      "loss": 0.8476,
+      "step": 9186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.485562370502279e-06,
+      "loss": 0.8453,
+      "step": 9187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4837515449626405e-06,
+      "loss": 0.8459,
+      "step": 9188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481941285752616e-06,
+      "loss": 0.8434,
+      "step": 9189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4801315930086147e-06,
+      "loss": 0.8761,
+      "step": 9190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478322466866985e-06,
+      "loss": 0.8855,
+      "step": 9191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.476513907464041e-06,
+      "loss": 0.8851,
+      "step": 9192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.474705914936053e-06,
+      "loss": 0.8915,
+      "step": 9193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.47289848941925e-06,
+      "loss": 0.8786,
+      "step": 9194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4710916310498155e-06,
+      "loss": 0.8843,
+      "step": 9195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.469285339963892e-06,
+      "loss": 0.8722,
+      "step": 9196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4674796162975778e-06,
+      "loss": 0.8673,
+      "step": 9197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.465674460186931e-06,
+      "loss": 0.8789,
+      "step": 9198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4638698717679653e-06,
+      "loss": 0.8683,
+      "step": 9199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4620658511766504e-06,
+      "loss": 0.8734,
+      "step": 9200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.460262398548916e-06,
+      "loss": 0.8485,
+      "step": 9201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4584595140206457e-06,
+      "loss": 0.8849,
+      "step": 9202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4566571977276845e-06,
+      "loss": 0.8559,
+      "step": 9203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4548554498058295e-06,
+      "loss": 0.8363,
+      "step": 9204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.45305427039084e-06,
+      "loss": 0.862,
+      "step": 9205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4512536596184288e-06,
+      "loss": 0.8397,
+      "step": 9206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4494536176242677e-06,
+      "loss": 0.9081,
+      "step": 9207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.447654144543986e-06,
+      "loss": 0.8309,
+      "step": 9208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4458552405131673e-06,
+      "loss": 0.8708,
+      "step": 9209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.444056905667356e-06,
+      "loss": 0.8434,
+      "step": 9210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4422591401420537e-06,
+      "loss": 0.8161,
+      "step": 9211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4404619440727095e-06,
+      "loss": 0.8886,
+      "step": 9212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4386653175947483e-06,
+      "loss": 0.8539,
+      "step": 9213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4368692608435294e-06,
+      "loss": 0.8767,
+      "step": 9214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4350737739543938e-06,
+      "loss": 0.8536,
+      "step": 9215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4332788570626143e-06,
+      "loss": 0.8558,
+      "step": 9216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4314845103034456e-06,
+      "loss": 0.8725,
+      "step": 9217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4296907338120733e-06,
+      "loss": 0.8504,
+      "step": 9218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4278975277236682e-06,
+      "loss": 0.8425,
+      "step": 9219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42610489217333e-06,
+      "loss": 0.8317,
+      "step": 9220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4243128272961412e-06,
+      "loss": 0.8856,
+      "step": 9221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4225213332271203e-06,
+      "loss": 0.8066,
+      "step": 9222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.420730410101255e-06,
+      "loss": 0.8867,
+      "step": 9223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4189400580534873e-06,
+      "loss": 0.8388,
+      "step": 9224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4171502772187148e-06,
+      "loss": 0.8316,
+      "step": 9225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.415361067731793e-06,
+      "loss": 0.8394,
+      "step": 9226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.413572429727533e-06,
+      "loss": 0.8907,
+      "step": 9227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4117843633407055e-06,
+      "loss": 0.8999,
+      "step": 9228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409996868706036e-06,
+      "loss": 0.8224,
+      "step": 9229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4082099459582078e-06,
+      "loss": 0.8768,
+      "step": 9230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.406423595231857e-06,
+      "loss": 0.8503,
+      "step": 9231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.40463781666159e-06,
+      "loss": 0.9101,
+      "step": 9232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4028526103819527e-06,
+      "loss": 0.8817,
+      "step": 9233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.401067976527457e-06,
+      "loss": 0.8463,
+      "step": 9234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.399283915232571e-06,
+      "loss": 0.8551,
+      "step": 9235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3975004266317193e-06,
+      "loss": 0.8358,
+      "step": 9236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3957175108592836e-06,
+      "loss": 0.8578,
+      "step": 9237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3939351680495994e-06,
+      "loss": 0.8925,
+      "step": 9238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.392153398336964e-06,
+      "loss": 0.8706,
+      "step": 9239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3903722018556285e-06,
+      "loss": 0.7699,
+      "step": 9240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3885915787398016e-06,
+      "loss": 0.873,
+      "step": 9241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3868115291236515e-06,
+      "loss": 0.8712,
+      "step": 9242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3850320531412908e-06,
+      "loss": 0.8647,
+      "step": 9243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3832531509268076e-06,
+      "loss": 0.8604,
+      "step": 9244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3814748226142335e-06,
+      "loss": 0.8531,
+      "step": 9245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.379697068337563e-06,
+      "loss": 0.8572,
+      "step": 9246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.379697068337563e-06,
+      "loss": 0.8201,
+      "step": 9247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3779198882307443e-06,
+      "loss": 0.8194,
+      "step": 9248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.376143282427683e-06,
+      "loss": 0.8142,
+      "step": 9249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3743672510622417e-06,
+      "loss": 0.885,
+      "step": 9250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3725917942682397e-06,
+      "loss": 0.8328,
+      "step": 9251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.370816912179452e-06,
+      "loss": 0.8381,
+      "step": 9252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3690426049296145e-06,
+      "loss": 0.8405,
+      "step": 9253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.367268872652416e-06,
+      "loss": 0.867,
+      "step": 9254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.365495715481495e-06,
+      "loss": 0.9006,
+      "step": 9255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3637231335504675e-06,
+      "loss": 0.8757,
+      "step": 9256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3619511269928784e-06,
+      "loss": 0.8322,
+      "step": 9257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3601796959422585e-06,
+      "loss": 0.5626,
+      "step": 9258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.358408840532066e-06,
+      "loss": 0.8979,
+      "step": 9259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3566385608957443e-06,
+      "loss": 0.8803,
+      "step": 9260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3548688571666657e-06,
+      "loss": 0.8635,
+      "step": 9261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3530997294781854e-06,
+      "loss": 0.7967,
+      "step": 9262
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3513311779635904e-06,
+      "loss": 0.8347,
+      "step": 9263
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.349563202756149e-06,
+      "loss": 0.8433,
+      "step": 9264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3477958039890638e-06,
+      "loss": 0.8775,
+      "step": 9265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3460289817955063e-06,
+      "loss": 0.9192,
+      "step": 9266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3442627363086026e-06,
+      "loss": 0.8972,
+      "step": 9267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3424970676614354e-06,
+      "loss": 0.8927,
+      "step": 9268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.340731975987042e-06,
+      "loss": 0.8577,
+      "step": 9269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3389674614184177e-06,
+      "loss": 0.908,
+      "step": 9270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.337203524088514e-06,
+      "loss": 0.9048,
+      "step": 9271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3354401641302395e-06,
+      "loss": 0.8745,
+      "step": 9272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.333677381676458e-06,
+      "loss": 0.9067,
+      "step": 9273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.33191517685999e-06,
+      "loss": 0.8788,
+      "step": 9274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.330153549813615e-06,
+      "loss": 0.9177,
+      "step": 9275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3283925006700657e-06,
+      "loss": 0.8628,
+      "step": 9276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.326632029562034e-06,
+      "loss": 0.8522,
+      "step": 9277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.324872136622164e-06,
+      "loss": 0.8654,
+      "step": 9278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323112821983061e-06,
+      "loss": 0.8654,
+      "step": 9279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.321354085777284e-06,
+      "loss": 0.8796,
+      "step": 9280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.319595928137349e-06,
+      "loss": 0.8502,
+      "step": 9281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31783834919573e-06,
+      "loss": 0.8767,
+      "step": 9282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3160813490848542e-06,
+      "loss": 0.8372,
+      "step": 9283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3143249279371085e-06,
+      "loss": 0.8552,
+      "step": 9284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.312569085884836e-06,
+      "loss": 0.8793,
+      "step": 9285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.310813823060327e-06,
+      "loss": 0.874,
+      "step": 9286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3090591395958485e-06,
+      "loss": 0.8298,
+      "step": 9287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.307305035623597e-06,
+      "loss": 0.8493,
+      "step": 9288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.305551511275751e-06,
+      "loss": 0.8472,
+      "step": 9289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3037985666844297e-06,
+      "loss": 0.8627,
+      "step": 9290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3020462019817137e-06,
+      "loss": 0.8758,
+      "step": 9291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3002944172996377e-06,
+      "loss": 0.8515,
+      "step": 9292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2985432127701945e-06,
+      "loss": 0.8826,
+      "step": 9293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.296792588525335e-06,
+      "loss": 0.8658,
+      "step": 9294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2950425446969605e-06,
+      "loss": 0.8664,
+      "step": 9295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2932930814169383e-06,
+      "loss": 0.8039,
+      "step": 9296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.291544198817075e-06,
+      "loss": 0.8493,
+      "step": 9297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2897958970291578e-06,
+      "loss": 0.9351,
+      "step": 9298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2880481761849037e-06,
+      "loss": 0.8635,
+      "step": 9299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.28630103641601e-06,
+      "loss": 0.8265,
+      "step": 9300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2845544778541095e-06,
+      "loss": 0.8768,
+      "step": 9301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.28280850063081e-06,
+      "loss": 0.8533,
+      "step": 9302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2810631048776566e-06,
+      "loss": 0.8344,
+      "step": 9303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.279318290726171e-06,
+      "loss": 0.848,
+      "step": 9304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.27757405830781e-06,
+      "loss": 0.8433,
+      "step": 9305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.275830407754006e-06,
+      "loss": 0.8221,
+      "step": 9306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2740873391961325e-06,
+      "loss": 0.8431,
+      "step": 9307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2723448527655267e-06,
+      "loss": 0.8532,
+      "step": 9308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2706029485934813e-06,
+      "loss": 0.8809,
+      "step": 9309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.268861626811242e-06,
+      "loss": 0.8912,
+      "step": 9310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.267120887550015e-06,
+      "loss": 0.8512,
+      "step": 9311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2653807309409593e-06,
+      "loss": 0.8797,
+      "step": 9312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.263641157115193e-06,
+      "loss": 0.8876,
+      "step": 9313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2619021662037855e-06,
+      "loss": 0.8546,
+      "step": 9314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.260163758337768e-06,
+      "loss": 0.8512,
+      "step": 9315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.258425933648122e-06,
+      "loss": 0.8806,
+      "step": 9316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2566886922657917e-06,
+      "loss": 0.8226,
+      "step": 9317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2549520343216704e-06,
+      "loss": 0.861,
+      "step": 9318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.253215959946612e-06,
+      "loss": 0.8422,
+      "step": 9319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2514804692714264e-06,
+      "loss": 0.9124,
+      "step": 9320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249745562426876e-06,
+      "loss": 0.8655,
+      "step": 9321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2480112395436826e-06,
+      "loss": 0.8759,
+      "step": 9322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.246277500752524e-06,
+      "loss": 0.847,
+      "step": 9323
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.244544346184031e-06,
+      "loss": 0.8504,
+      "step": 9324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2428117759687926e-06,
+      "loss": 0.8939,
+      "step": 9325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.241079790237355e-06,
+      "loss": 0.8636,
+      "step": 9326
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2393483891202194e-06,
+      "loss": 0.8372,
+      "step": 9327
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2376175727478346e-06,
+      "loss": 0.8642,
+      "step": 9328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2358873412506254e-06,
+      "loss": 0.8385,
+      "step": 9329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2341576947589472e-06,
+      "loss": 0.8727,
+      "step": 9330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2324286334031377e-06,
+      "loss": 0.8691,
+      "step": 9331
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2307001573134646e-06,
+      "loss": 0.8286,
+      "step": 9332
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.228972266620172e-06,
+      "loss": 0.8459,
+      "step": 9333
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.227244961453451e-06,
+      "loss": 0.8314,
+      "step": 9334
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.225518241943446e-06,
+      "loss": 0.8644,
+      "step": 9335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.223792108220265e-06,
+      "loss": 0.8849,
+      "step": 9336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.222066560413966e-06,
+      "loss": 0.8688,
+      "step": 9337
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.220341598654565e-06,
+      "loss": 0.831,
+      "step": 9338
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2186172230720293e-06,
+      "loss": 0.8404,
+      "step": 9339
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2168934337962932e-06,
+      "loss": 0.8632,
+      "step": 9340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.215170230957231e-06,
+      "loss": 0.8193,
+      "step": 9341
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.213447614684692e-06,
+      "loss": 0.8948,
+      "step": 9342
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2117255851084595e-06,
+      "loss": 0.8818,
+      "step": 9343
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2100041423582954e-06,
+      "loss": 0.9086,
+      "step": 9344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.208283286563894e-06,
+      "loss": 0.8367,
+      "step": 9345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.206563017854929e-06,
+      "loss": 0.4934,
+      "step": 9346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2048433363610077e-06,
+      "loss": 0.8302,
+      "step": 9347
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2031242422117127e-06,
+      "loss": 0.863,
+      "step": 9348
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.201405735536567e-06,
+      "loss": 0.8837,
+      "step": 9349
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.199687816465058e-06,
+      "loss": 0.8611,
+      "step": 9350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.197970485126625e-06,
+      "loss": 0.8373,
+      "step": 9351
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1962537416506656e-06,
+      "loss": 0.8337,
+      "step": 9352
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.194537586166532e-06,
+      "loss": 0.5507,
+      "step": 9353
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.192822018803532e-06,
+      "loss": 0.8452,
+      "step": 9354
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1911070396909285e-06,
+      "loss": 0.8456,
+      "step": 9355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18939264895794e-06,
+      "loss": 0.8745,
+      "step": 9356
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1876788467337427e-06,
+      "loss": 0.8868,
+      "step": 9357
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1859656331474677e-06,
+      "loss": 0.8048,
+      "step": 9358
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.184253008328199e-06,
+      "loss": 0.8183,
+      "step": 9359
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1825409724049807e-06,
+      "loss": 0.8656,
+      "step": 9360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.180829525506808e-06,
+      "loss": 0.8588,
+      "step": 9361
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.179118667762635e-06,
+      "loss": 0.8396,
+      "step": 9362
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1774083993013715e-06,
+      "loss": 0.8115,
+      "step": 9363
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.17569872025188e-06,
+      "loss": 0.9075,
+      "step": 9364
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.173989630742981e-06,
+      "loss": 0.8519,
+      "step": 9365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1722811309034486e-06,
+      "loss": 0.8486,
+      "step": 9366
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.170573220862017e-06,
+      "loss": 0.8336,
+      "step": 9367
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16886590074737e-06,
+      "loss": 0.8716,
+      "step": 9368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.167159170688152e-06,
+      "loss": 0.8316,
+      "step": 9369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165453030812955e-06,
+      "loss": 0.8574,
+      "step": 9370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163747481250342e-06,
+      "loss": 0.8837,
+      "step": 9371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.16204252212881e-06,
+      "loss": 0.8485,
+      "step": 9372
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.160338153576834e-06,
+      "loss": 0.8793,
+      "step": 9373
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1586343757228247e-06,
+      "loss": 0.8671,
+      "step": 9374
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.156931188695166e-06,
+      "loss": 0.861,
+      "step": 9375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1552285926221792e-06,
+      "loss": 0.8868,
+      "step": 9376
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1535265876321574e-06,
+      "loss": 0.8601,
+      "step": 9377
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1518251738533413e-06,
+      "loss": 0.8279,
+      "step": 9378
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.150124351413926e-06,
+      "loss": 0.8724,
+      "step": 9379
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1484241204420676e-06,
+      "loss": 0.8508,
+      "step": 9380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1467244810658648e-06,
+      "loss": 0.5023,
+      "step": 9381
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1450254334133945e-06,
+      "loss": 0.8575,
+      "step": 9382
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.143326977612662e-06,
+      "loss": 0.8245,
+      "step": 9383
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1416291137916535e-06,
+      "loss": 0.8461,
+      "step": 9384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1399318420782876e-06,
+      "loss": 0.9048,
+      "step": 9385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1382351626004595e-06,
+      "loss": 0.8694,
+      "step": 9386
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1365390754860006e-06,
+      "loss": 0.8305,
+      "step": 9387
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1348435808627153e-06,
+      "loss": 0.8251,
+      "step": 9388
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1331486788583444e-06,
+      "loss": 0.8813,
+      "step": 9389
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1314543696006063e-06,
+      "loss": 0.8518,
+      "step": 9390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1297606532171532e-06,
+      "loss": 0.8849,
+      "step": 9391
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.128067529835606e-06,
+      "loss": 0.8413,
+      "step": 9392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.126374999583536e-06,
+      "loss": 0.8539,
+      "step": 9393
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1246830625884706e-06,
+      "loss": 0.8767,
+      "step": 9394
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122991718977896e-06,
+      "loss": 0.8547,
+      "step": 9395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1213009688792465e-06,
+      "loss": 0.8791,
+      "step": 9396
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1196108124199177e-06,
+      "loss": 0.8157,
+      "step": 9397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1179212497272582e-06,
+      "loss": 0.8434,
+      "step": 9398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.116232280928573e-06,
+      "loss": 0.8455,
+      "step": 9399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1145439061511196e-06,
+      "loss": 0.875,
+      "step": 9400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1128561255221138e-06,
+      "loss": 0.8605,
+      "step": 9401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.111168939168725e-06,
+      "loss": 0.839,
+      "step": 9402
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.109482347218079e-06,
+      "loss": 0.509,
+      "step": 9403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1077963497972555e-06,
+      "loss": 0.8174,
+      "step": 9404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.106110947033291e-06,
+      "loss": 0.8351,
+      "step": 9405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.104426139053174e-06,
+      "loss": 0.8621,
+      "step": 9406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.102741925983851e-06,
+      "loss": 0.4999,
+      "step": 9407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1010583079522253e-06,
+      "loss": 0.8654,
+      "step": 9408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09937528508515e-06,
+      "loss": 0.8187,
+      "step": 9409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.097692857509439e-06,
+      "loss": 0.8669,
+      "step": 9410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0960110253518596e-06,
+      "loss": 0.8517,
+      "step": 9411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.094329788739127e-06,
+      "loss": 0.8115,
+      "step": 9412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0926491477979272e-06,
+      "loss": 0.5687,
+      "step": 9413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.090969102654883e-06,
+      "loss": 0.896,
+      "step": 9414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08928965343659e-06,
+      "loss": 0.8234,
+      "step": 9415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0876108002695817e-06,
+      "loss": 0.8404,
+      "step": 9416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0859325432803647e-06,
+      "loss": 0.8409,
+      "step": 9417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0842548825953813e-06,
+      "loss": 0.5021,
+      "step": 9418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0825778183410485e-06,
+      "loss": 0.8618,
+      "step": 9419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.080901350643719e-06,
+      "loss": 0.8386,
+      "step": 9420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0792254796297175e-06,
+      "loss": 0.859,
+      "step": 9421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.077550205425317e-06,
+      "loss": 0.8684,
+      "step": 9422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0758755281567376e-06,
+      "loss": 0.8906,
+      "step": 9423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0742014479501717e-06,
+      "loss": 0.8334,
+      "step": 9424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0725279649317463e-06,
+      "loss": 0.8035,
+      "step": 9425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.070855079227565e-06,
+      "loss": 0.8368,
+      "step": 9426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.069182790963663e-06,
+      "loss": 0.8792,
+      "step": 9427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0675111002660566e-06,
+      "loss": 0.8766,
+      "step": 9428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.06584000726069e-06,
+      "loss": 0.8713,
+      "step": 9429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064169512073486e-06,
+      "loss": 0.8466,
+      "step": 9430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0624996148303043e-06,
+      "loss": 0.8469,
+      "step": 9431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0608303156569763e-06,
+      "loss": 0.8837,
+      "step": 9432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0591616146792705e-06,
+      "loss": 0.8625,
+      "step": 9433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0574935120229224e-06,
+      "loss": 0.9153,
+      "step": 9434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0558260078136203e-06,
+      "loss": 0.8836,
+      "step": 9435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0541591021770047e-06,
+      "loss": 0.8764,
+      "step": 9436
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.052492795238673e-06,
+      "loss": 0.8755,
+      "step": 9437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0508270871241776e-06,
+      "loss": 0.7868,
+      "step": 9438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0491619779590245e-06,
+      "loss": 0.8798,
+      "step": 9439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0474974678686764e-06,
+      "loss": 0.894,
+      "step": 9440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0458335569785493e-06,
+      "loss": 0.8485,
+      "step": 9441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0441702454140134e-06,
+      "loss": 0.8534,
+      "step": 9442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.042507533300395e-06,
+      "loss": 0.8321,
+      "step": 9443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.040845420762978e-06,
+      "loss": 0.9121,
+      "step": 9444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.039183907926995e-06,
+      "loss": 0.8532,
+      "step": 9445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0375229949176367e-06,
+      "loss": 0.8362,
+      "step": 9446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0358626818600513e-06,
+      "loss": 0.8616,
+      "step": 9447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0342029688793364e-06,
+      "loss": 0.8565,
+      "step": 9448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.032543856100548e-06,
+      "loss": 0.864,
+      "step": 9449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0308853436486953e-06,
+      "loss": 0.8275,
+      "step": 9450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0292274316487447e-06,
+      "loss": 0.8608,
+      "step": 9451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027570120225614e-06,
+      "loss": 0.866,
+      "step": 9452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0259134095041788e-06,
+      "loss": 0.8945,
+      "step": 9453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0242572996092624e-06,
+      "loss": 0.8515,
+      "step": 9454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0226017906656568e-06,
+      "loss": 0.8664,
+      "step": 9455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0209468827980914e-06,
+      "loss": 0.8429,
+      "step": 9456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.01929257613127e-06,
+      "loss": 0.8364,
+      "step": 9457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0176388707898274e-06,
+      "loss": 0.867,
+      "step": 9458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0159857668983785e-06,
+      "loss": 0.8846,
+      "step": 9459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0143332645814685e-06,
+      "loss": 0.8683,
+      "step": 9460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.012681363963621e-06,
+      "loss": 0.7971,
+      "step": 9461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0110300651692905e-06,
+      "loss": 0.8498,
+      "step": 9462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.00937936832291e-06,
+      "loss": 0.8364,
+      "step": 9463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0077292735488465e-06,
+      "loss": 0.8108,
+      "step": 9464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0060797809714296e-06,
+      "loss": 0.9041,
+      "step": 9465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0044308907149533e-06,
+      "loss": 0.8422,
+      "step": 9466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.002782602903647e-06,
+      "loss": 0.8547,
+      "step": 9467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0011349176617133e-06,
+      "loss": 0.8261,
+      "step": 9468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9994878351132916e-06,
+      "loss": 0.8124,
+      "step": 9469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997841355382497e-06,
+      "loss": 0.9139,
+      "step": 9470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9961954785933758e-06,
+      "loss": 0.8339,
+      "step": 9471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9945502048699504e-06,
+      "loss": 0.53,
+      "step": 9472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.992905534336178e-06,
+      "loss": 0.8135,
+      "step": 9473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9912614671159914e-06,
+      "loss": 0.8413,
+      "step": 9474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.989618003333258e-06,
+      "loss": 0.8989,
+      "step": 9475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98797514311181e-06,
+      "loss": 0.8598,
+      "step": 9476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986332886575435e-06,
+      "loss": 0.8326,
+      "step": 9477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.984691233847872e-06,
+      "loss": 0.8229,
+      "step": 9478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983050185052815e-06,
+      "loss": 0.8173,
+      "step": 9479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.981409740313912e-06,
+      "loss": 0.84,
+      "step": 9480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9797698997547663e-06,
+      "loss": 0.8333,
+      "step": 9481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9781306634989373e-06,
+      "loss": 0.5035,
+      "step": 9482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9764920316699353e-06,
+      "loss": 0.8391,
+      "step": 9483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.974854004391228e-06,
+      "loss": 0.8565,
+      "step": 9484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.973216581786235e-06,
+      "loss": 0.8283,
+      "step": 9485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.971579763978333e-06,
+      "loss": 0.8439,
+      "step": 9486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.969943551090853e-06,
+      "loss": 0.8625,
+      "step": 9487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9683079432470774e-06,
+      "loss": 0.8956,
+      "step": 9488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9666729405702457e-06,
+      "loss": 0.8591,
+      "step": 9489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9650385431835508e-06,
+      "loss": 0.8855,
+      "step": 9490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9634047512101405e-06,
+      "loss": 0.8292,
+      "step": 9491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9617715647731172e-06,
+      "loss": 0.8149,
+      "step": 9492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9601389839955366e-06,
+      "loss": 0.812,
+      "step": 9493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.958507009000409e-06,
+      "loss": 0.8283,
+      "step": 9494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9568756399107037e-06,
+      "loss": 0.8428,
+      "step": 9495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9552448768493293e-06,
+      "loss": 0.8621,
+      "step": 9496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.953614719939173e-06,
+      "loss": 0.795,
+      "step": 9497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951985169303051e-06,
+      "loss": 0.8619,
+      "step": 9498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9503562250637574e-06,
+      "loss": 0.8855,
+      "step": 9499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.948727887344016e-06,
+      "loss": 0.8822,
+      "step": 9500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9471001562665305e-06,
+      "loss": 0.8287,
+      "step": 9501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.945473031953934e-06,
+      "loss": 0.8771,
+      "step": 9502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9438465145288377e-06,
+      "loss": 0.8688,
+      "step": 9503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.942220604113785e-06,
+      "loss": 0.8604,
+      "step": 9504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.940595300831294e-06,
+      "loss": 0.7908,
+      "step": 9505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9389706048038183e-06,
+      "loss": 0.8128,
+      "step": 9506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9373465161537787e-06,
+      "loss": 0.7951,
+      "step": 9507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.935723035003545e-06,
+      "loss": 0.8588,
+      "step": 9508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9341001614754407e-06,
+      "loss": 0.8754,
+      "step": 9509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.932477895691751e-06,
+      "loss": 0.8378,
+      "step": 9510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.930856237774701e-06,
+      "loss": 0.861,
+      "step": 9511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.929235187846488e-06,
+      "loss": 0.8773,
+      "step": 9512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9276147460292437e-06,
+      "loss": 0.8569,
+      "step": 9513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9259949124450726e-06,
+      "loss": 0.8309,
+      "step": 9514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.924375687216018e-06,
+      "loss": 0.874,
+      "step": 9515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.922757070464092e-06,
+      "loss": 0.8752,
+      "step": 9516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.921139062311246e-06,
+      "loss": 0.8245,
+      "step": 9517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9195216628793956e-06,
+      "loss": 0.8721,
+      "step": 9518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.917904872290407e-06,
+      "loss": 0.8431,
+      "step": 9519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9162886906661025e-06,
+      "loss": 0.8614,
+      "step": 9520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9146731181282554e-06,
+      "loss": 0.8383,
+      "step": 9521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.913058154798596e-06,
+      "loss": 0.8392,
+      "step": 9522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.911443800798808e-06,
+      "loss": 0.7985,
+      "step": 9523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.8492,
+      "step": 9524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9082169212753455e-06,
+      "loss": 0.9,
+      "step": 9525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.906604395994809e-06,
+      "loss": 0.8145,
+      "step": 9526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9049924805304165e-06,
+      "loss": 0.8999,
+      "step": 9527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9033811750036224e-06,
+      "loss": 0.8869,
+      "step": 9528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9017704795358338e-06,
+      "loss": 0.8258,
+      "step": 9529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9001603942484127e-06,
+      "loss": 0.8828,
+      "step": 9530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8985509192626739e-06,
+      "loss": 0.8337,
+      "step": 9531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8969420546998896e-06,
+      "loss": 0.8946,
+      "step": 9532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8953338006812805e-06,
+      "loss": 0.886,
+      "step": 9533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8937261573280252e-06,
+      "loss": 0.8814,
+      "step": 9534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8921191247612569e-06,
+      "loss": 0.7935,
+      "step": 9535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8905127031020598e-06,
+      "loss": 0.7945,
+      "step": 9536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.888906892471477e-06,
+      "loss": 0.7944,
+      "step": 9537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8873016929904942e-06,
+      "loss": 0.8757,
+      "step": 9538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8856971047800687e-06,
+      "loss": 0.8338,
+      "step": 9539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.884093127961092e-06,
+      "loss": 0.8629,
+      "step": 9540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8824897626544292e-06,
+      "loss": 0.8349,
+      "step": 9541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8808870089808806e-06,
+      "loss": 0.8209,
+      "step": 9542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8792848670612195e-06,
+      "loss": 0.8923,
+      "step": 9543
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.877683337016153e-06,
+      "loss": 0.8687,
+      "step": 9544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8760824189663618e-06,
+      "loss": 0.8423,
+      "step": 9545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8744821130324609e-06,
+      "loss": 0.861,
+      "step": 9546
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8728824193350381e-06,
+      "loss": 0.8325,
+      "step": 9547
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8712833379946217e-06,
+      "loss": 0.864,
+      "step": 9548
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.869684869131697e-06,
+      "loss": 0.8253,
+      "step": 9549
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8680870128667073e-06,
+      "loss": 0.8774,
+      "step": 9550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8664897693200456e-06,
+      "loss": 0.8617,
+      "step": 9551
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8648931386120605e-06,
+      "loss": 0.8539,
+      "step": 9552
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8632971208630513e-06,
+      "loss": 0.8793,
+      "step": 9553
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8617017161932815e-06,
+      "loss": 0.8421,
+      "step": 9554
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8601069247229497e-06,
+      "loss": 0.8415,
+      "step": 9555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8585127465722308e-06,
+      "loss": 0.8419,
+      "step": 9556
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8569191818612298e-06,
+      "loss": 0.8223,
+      "step": 9557
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8553262307100295e-06,
+      "loss": 0.8518,
+      "step": 9558
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8537338932386462e-06,
+      "loss": 0.8428,
+      "step": 9559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8521421695670617e-06,
+      "loss": 0.8377,
+      "step": 9560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8505510598152077e-06,
+      "loss": 0.8402,
+      "step": 9561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8489605641029694e-06,
+      "loss": 0.8336,
+      "step": 9562
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.847370682550187e-06,
+      "loss": 0.8778,
+      "step": 9563
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8457814152766551e-06,
+      "loss": 0.8795,
+      "step": 9564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8441927624021194e-06,
+      "loss": 0.8554,
+      "step": 9565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8426047240462807e-06,
+      "loss": 0.8859,
+      "step": 9566
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8410173003287935e-06,
+      "loss": 0.896,
+      "step": 9567
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8394304913692673e-06,
+      "loss": 0.8334,
+      "step": 9568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8378442972872635e-06,
+      "loss": 0.843,
+      "step": 9569
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.836258718202296e-06,
+      "loss": 0.838,
+      "step": 9570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8346737542338366e-06,
+      "loss": 0.8427,
+      "step": 9571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.833089405501307e-06,
+      "loss": 0.4789,
+      "step": 9572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8315056721240831e-06,
+      "loss": 0.8755,
+      "step": 9573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.829922554221496e-06,
+      "loss": 0.8486,
+      "step": 9574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.828340051912828e-06,
+      "loss": 0.8588,
+      "step": 9575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8267581653173195e-06,
+      "loss": 0.8399,
+      "step": 9576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8251768945541581e-06,
+      "loss": 0.8336,
+      "step": 9577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.82359623974249e-06,
+      "loss": 0.8846,
+      "step": 9578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8220162010014163e-06,
+      "loss": 0.8214,
+      "step": 9579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.820436778449981e-06,
+      "loss": 0.8678,
+      "step": 9580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8188579722071985e-06,
+      "loss": 0.8739,
+      "step": 9581
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8172797823920196e-06,
+      "loss": 0.8624,
+      "step": 9582
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8157022091233645e-06,
+      "loss": 0.8686,
+      "step": 9583
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8141252525200914e-06,
+      "loss": 0.8658,
+      "step": 9584
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8125489127010275e-06,
+      "loss": 0.8215,
+      "step": 9585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8109731897849382e-06,
+      "loss": 0.8356,
+      "step": 9586
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.809398083890559e-06,
+      "loss": 0.8473,
+      "step": 9587
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8078235951365597e-06,
+      "loss": 0.8668,
+      "step": 9588
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8062497236415832e-06,
+      "loss": 0.8972,
+      "step": 9589
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8046764695242115e-06,
+      "loss": 0.8298,
+      "step": 9590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.803103832902986e-06,
+      "loss": 0.882,
+      "step": 9591
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8015318138964011e-06,
+      "loss": 0.8668,
+      "step": 9592
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7999604126229043e-06,
+      "loss": 0.8708,
+      "step": 9593
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7983896292008963e-06,
+      "loss": 0.8798,
+      "step": 9594
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7968194637487323e-06,
+      "loss": 0.8437,
+      "step": 9595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7952499163847192e-06,
+      "loss": 0.8665,
+      "step": 9596
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7936809872271167e-06,
+      "loss": 0.8731,
+      "step": 9597
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7921126763941454e-06,
+      "loss": 0.8743,
+      "step": 9598
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7905449840039645e-06,
+      "loss": 0.8744,
+      "step": 9599
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7889779101747063e-06,
+      "loss": 0.828,
+      "step": 9600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7874114550244381e-06,
+      "loss": 0.8393,
+      "step": 9601
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.785845618671188e-06,
+      "loss": 0.8089,
+      "step": 9602
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7842804012329419e-06,
+      "loss": 0.8795,
+      "step": 9603
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7827158028276315e-06,
+      "loss": 0.8836,
+      "step": 9604
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7811518235731461e-06,
+      "loss": 0.8597,
+      "step": 9605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7795884635873272e-06,
+      "loss": 0.8377,
+      "step": 9606
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7780257229879717e-06,
+      "loss": 0.8706,
+      "step": 9607
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7764636018928249e-06,
+      "loss": 0.8305,
+      "step": 9608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.774902100419591e-06,
+      "loss": 0.8369,
+      "step": 9609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7733412186859244e-06,
+      "loss": 0.8204,
+      "step": 9610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7717809568094334e-06,
+      "loss": 0.9114,
+      "step": 9611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7702213149076797e-06,
+      "loss": 0.8272,
+      "step": 9612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.768662293098178e-06,
+      "loss": 0.8264,
+      "step": 9613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7671038914983963e-06,
+      "loss": 0.5097,
+      "step": 9614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7655461102257566e-06,
+      "loss": 0.8695,
+      "step": 9615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7639889493976337e-06,
+      "loss": 0.8614,
+      "step": 9616
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.762432409131355e-06,
+      "loss": 0.854,
+      "step": 9617
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.760876489544202e-06,
+      "loss": 0.8877,
+      "step": 9618
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7593211907534092e-06,
+      "loss": 0.8903,
+      "step": 9619
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7577665128761645e-06,
+      "loss": 0.8883,
+      "step": 9620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7562124560296113e-06,
+      "loss": 0.8685,
+      "step": 9621
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7546590203308355e-06,
+      "loss": 0.8548,
+      "step": 9622
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.753106205896895e-06,
+      "loss": 0.8779,
+      "step": 9623
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7515540128447793e-06,
+      "loss": 0.909,
+      "step": 9624
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7500024412914541e-06,
+      "loss": 0.8992,
+      "step": 9625
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7484514913538154e-06,
+      "loss": 0.8049,
+      "step": 9626
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.746901163148732e-06,
+      "loss": 0.8334,
+      "step": 9627
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7453514567930085e-06,
+      "loss": 0.8688,
+      "step": 9628
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7438023724034215e-06,
+      "loss": 0.8598,
+      "step": 9629
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7422539100966784e-06,
+      "loss": 0.8709,
+      "step": 9630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.740706069989464e-06,
+      "loss": 0.8536,
+      "step": 9631
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7391588521983948e-06,
+      "loss": 0.8308,
+      "step": 9632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7376122568400533e-06,
+      "loss": 0.8503,
+      "step": 9633
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7360662840309706e-06,
+      "loss": 0.8172,
+      "step": 9634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7345209338876324e-06,
+      "loss": 0.8343,
+      "step": 9635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7329762065264755e-06,
+      "loss": 0.8539,
+      "step": 9636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7314321020638924e-06,
+      "loss": 0.8702,
+      "step": 9637
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.729888620616228e-06,
+      "loss": 0.8471,
+      "step": 9638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.728345762299778e-06,
+      "loss": 0.8612,
+      "step": 9639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7268035272307927e-06,
+      "loss": 0.8515,
+      "step": 9640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7252619155254734e-06,
+      "loss": 0.9087,
+      "step": 9641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7237209272999845e-06,
+      "loss": 0.8585,
+      "step": 9642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.722180562670428e-06,
+      "loss": 0.8795,
+      "step": 9643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7206408217528669e-06,
+      "loss": 0.8386,
+      "step": 9644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7191017046633196e-06,
+      "loss": 0.8605,
+      "step": 9645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.717563211517752e-06,
+      "loss": 0.8376,
+      "step": 9646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7160253424320872e-06,
+      "loss": 0.8451,
+      "step": 9647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.714488097522199e-06,
+      "loss": 0.8757,
+      "step": 9648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7129514769039147e-06,
+      "loss": 0.8621,
+      "step": 9649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7114154806930138e-06,
+      "loss": 0.86,
+      "step": 9650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709880109005232e-06,
+      "loss": 0.8292,
+      "step": 9651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708345361956255e-06,
+      "loss": 0.8691,
+      "step": 9652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7068112396617164e-06,
+      "loss": 0.7977,
+      "step": 9653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7052777422372158e-06,
+      "loss": 0.8599,
+      "step": 9654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703744869798295e-06,
+      "loss": 0.9008,
+      "step": 9655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7022126224604529e-06,
+      "loss": 0.7973,
+      "step": 9656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7006810003391382e-06,
+      "loss": 0.8901,
+      "step": 9657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6991500035497576e-06,
+      "loss": 0.838,
+      "step": 9658
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6976196322076655e-06,
+      "loss": 0.8715,
+      "step": 9659
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6960898864281717e-06,
+      "loss": 0.7712,
+      "step": 9660
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6945607663265384e-06,
+      "loss": 0.4822,
+      "step": 9661
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6930322720179816e-06,
+      "loss": 0.8609,
+      "step": 9662
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6915044036176719e-06,
+      "loss": 0.8609,
+      "step": 9663
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6899771612407223e-06,
+      "loss": 0.5878,
+      "step": 9664
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.688450545002216e-06,
+      "loss": 0.7945,
+      "step": 9665
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6869245550171697e-06,
+      "loss": 0.7928,
+      "step": 9666
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.685399191400573e-06,
+      "loss": 0.8616,
+      "step": 9667
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6838744542673492e-06,
+      "loss": 0.8431,
+      "step": 9668
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6823503437323908e-06,
+      "loss": 0.8642,
+      "step": 9669
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6808268599105282e-06,
+      "loss": 0.8817,
+      "step": 9670
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6793040029165596e-06,
+      "loss": 0.8204,
+      "step": 9671
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.677781772865219e-06,
+      "loss": 0.8469,
+      "step": 9672
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6762601698712134e-06,
+      "loss": 0.845,
+      "step": 9673
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.674739194049183e-06,
+      "loss": 0.8585,
+      "step": 9674
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6732188455137322e-06,
+      "loss": 0.5372,
+      "step": 9675
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6716991243794157e-06,
+      "loss": 0.8688,
+      "step": 9676
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6701800307607397e-06,
+      "loss": 0.8477,
+      "step": 9677
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6686615647721638e-06,
+      "loss": 0.8216,
+      "step": 9678
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.667143726528102e-06,
+      "loss": 0.8726,
+      "step": 9679
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6656265161429186e-06,
+      "loss": 0.5227,
+      "step": 9680
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6641099337309308e-06,
+      "loss": 0.8475,
+      "step": 9681
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6625939794064094e-06,
+      "loss": 0.8121,
+      "step": 9682
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6610786532835776e-06,
+      "loss": 0.8438,
+      "step": 9683
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6595639554766108e-06,
+      "loss": 0.9033,
+      "step": 9684
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6580498860996386e-06,
+      "loss": 0.8768,
+      "step": 9685
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.656536445266742e-06,
+      "loss": 0.8238,
+      "step": 9686
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6550236330919544e-06,
+      "loss": 0.8227,
+      "step": 9687
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6535114496892623e-06,
+      "loss": 0.8613,
+      "step": 9688
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6519998951726045e-06,
+      "loss": 0.8163,
+      "step": 9689
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6504889696558735e-06,
+      "loss": 0.8228,
+      "step": 9690
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6489786732529122e-06,
+      "loss": 0.87,
+      "step": 9691
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6474690060775178e-06,
+      "loss": 0.8803,
+      "step": 9692
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6459599682434413e-06,
+      "loss": 0.875,
+      "step": 9693
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.644451559864385e-06,
+      "loss": 0.8779,
+      "step": 9694
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6429437810539982e-06,
+      "loss": 0.8767,
+      "step": 9695
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6414366319258967e-06,
+      "loss": 0.8636,
+      "step": 9696
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6399301125936307e-06,
+      "loss": 0.8549,
+      "step": 9697
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6384242231707203e-06,
+      "loss": 0.8593,
+      "step": 9698
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6369189637706272e-06,
+      "loss": 0.8589,
+      "step": 9699
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6354143345067685e-06,
+      "loss": 0.8678,
+      "step": 9700
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6339103354925146e-06,
+      "loss": 0.5518,
+      "step": 9701
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6324069668411879e-06,
+      "loss": 0.8214,
+      "step": 9702
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6309042286660627e-06,
+      "loss": 0.8231,
+      "step": 9703
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6309042286660627e-06,
+      "loss": 0.8497,
+      "step": 9704
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.629402121080368e-06,
+      "loss": 0.8217,
+      "step": 9705
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6279006441972845e-06,
+      "loss": 0.8398,
+      "step": 9706
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6263997981299373e-06,
+      "loss": 0.8582,
+      "step": 9707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6248995829914216e-06,
+      "loss": 0.8236,
+      "step": 9708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6233999988947647e-06,
+      "loss": 0.8729,
+      "step": 9709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.621901045952966e-06,
+      "loss": 0.8252,
+      "step": 9710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6204027242789577e-06,
+      "loss": 0.4704,
+      "step": 9711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6189050339856448e-06,
+      "loss": 0.9042,
+      "step": 9712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.617407975185865e-06,
+      "loss": 0.8496,
+      "step": 9713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6159115479924259e-06,
+      "loss": 0.8529,
+      "step": 9714
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6144157525180715e-06,
+      "loss": 0.8475,
+      "step": 9715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.612920588875514e-06,
+      "loss": 0.83,
+      "step": 9716
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6114260571774031e-06,
+      "loss": 0.8117,
+      "step": 9717
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6099321575363513e-06,
+      "loss": 0.8936,
+      "step": 9718
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6084388900649195e-06,
+      "loss": 0.8264,
+      "step": 9719
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.606946254875621e-06,
+      "loss": 0.8222,
+      "step": 9720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6054542520809223e-06,
+      "loss": 0.8445,
+      "step": 9721
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.603962881793243e-06,
+      "loss": 0.83,
+      "step": 9722
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6024721441249525e-06,
+      "loss": 0.8418,
+      "step": 9723
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6009820391883746e-06,
+      "loss": 0.839,
+      "step": 9724
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5994925670957839e-06,
+      "loss": 0.9072,
+      "step": 9725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5980037279594097e-06,
+      "loss": 0.8543,
+      "step": 9726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5965155218914308e-06,
+      "loss": 0.8443,
+      "step": 9727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.59502794900398e-06,
+      "loss": 0.8671,
+      "step": 9728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.593541009409143e-06,
+      "loss": 0.8378,
+      "step": 9729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.592054703218956e-06,
+      "loss": 0.825,
+      "step": 9730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5905690305454068e-06,
+      "loss": 0.8714,
+      "step": 9731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5890839915004398e-06,
+      "loss": 0.8681,
+      "step": 9732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587599586195947e-06,
+      "loss": 0.8321,
+      "step": 9733
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5861158147437749e-06,
+      "loss": 0.8805,
+      "step": 9734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.584632677255723e-06,
+      "loss": 0.8495,
+      "step": 9735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5831501738435384e-06,
+      "loss": 0.8695,
+      "step": 9736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5816683046189296e-06,
+      "loss": 0.8159,
+      "step": 9737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.580187069693544e-06,
+      "loss": 0.8221,
+      "step": 9738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5787064691789966e-06,
+      "loss": 0.8733,
+      "step": 9739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5772265031868373e-06,
+      "loss": 0.8333,
+      "step": 9740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.575747171828589e-06,
+      "loss": 0.8915,
+      "step": 9741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5742684752157056e-06,
+      "loss": 0.4775,
+      "step": 9742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5727904134596084e-06,
+      "loss": 0.8496,
+      "step": 9743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5713129866716647e-06,
+      "loss": 0.8202,
+      "step": 9744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5698361949631946e-06,
+      "loss": 0.8509,
+      "step": 9745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.568360038445469e-06,
+      "loss": 0.878,
+      "step": 9746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5668845172297143e-06,
+      "loss": 0.8785,
+      "step": 9747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5654096314271093e-06,
+      "loss": 0.8449,
+      "step": 9748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5639353811487744e-06,
+      "loss": 0.8416,
+      "step": 9749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5624617665058005e-06,
+      "loss": 0.7672,
+      "step": 9750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5609887876092122e-06,
+      "loss": 0.9122,
+      "step": 9751
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5595164445700028e-06,
+      "loss": 0.8307,
+      "step": 9752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5580447374991003e-06,
+      "loss": 0.8325,
+      "step": 9753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.556573666507404e-06,
+      "loss": 0.8167,
+      "step": 9754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5551032317057456e-06,
+      "loss": 0.8657,
+      "step": 9755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5536334332049274e-06,
+      "loss": 0.8793,
+      "step": 9756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5521642711156859e-06,
+      "loss": 0.8592,
+      "step": 9757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.550695745548727e-06,
+      "loss": 0.8887,
+      "step": 9758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5492278566146945e-06,
+      "loss": 0.8841,
+      "step": 9759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5477606044241923e-06,
+      "loss": 0.832,
+      "step": 9760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5462939890877738e-06,
+      "loss": 0.8965,
+      "step": 9761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5448280107159442e-06,
+      "loss": 0.8902,
+      "step": 9762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5433626694191605e-06,
+      "loss": 0.8259,
+      "step": 9763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5418979653078348e-06,
+      "loss": 0.8514,
+      "step": 9764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.540433898492326e-06,
+      "loss": 0.8052,
+      "step": 9765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5389704690829488e-06,
+      "loss": 0.8709,
+      "step": 9766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.53750767718997e-06,
+      "loss": 0.8703,
+      "step": 9767
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5360455229236049e-06,
+      "loss": 0.4976,
+      "step": 9768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.534584006394023e-06,
+      "loss": 0.845,
+      "step": 9769
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.533123127711349e-06,
+      "loss": 0.8524,
+      "step": 9770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.531662886985652e-06,
+      "loss": 0.8594,
+      "step": 9771
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5302032843269609e-06,
+      "loss": 0.8694,
+      "step": 9772
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5287443198452502e-06,
+      "loss": 0.466,
+      "step": 9773
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5272859936504513e-06,
+      "loss": 0.8735,
+      "step": 9774
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5258283058524427e-06,
+      "loss": 0.8454,
+      "step": 9775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5243712565610603e-06,
+      "loss": 0.8572,
+      "step": 9776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5229148458860865e-06,
+      "loss": 0.8877,
+      "step": 9777
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5214590739372592e-06,
+      "loss": 0.8713,
+      "step": 9778
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5200039408242695e-06,
+      "loss": 0.8383,
+      "step": 9779
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5185494466567508e-06,
+      "loss": 0.8547,
+      "step": 9780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5170955915443043e-06,
+      "loss": 0.5358,
+      "step": 9781
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5156423755964645e-06,
+      "loss": 0.8545,
+      "step": 9782
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5141897989227372e-06,
+      "loss": 0.8154,
+      "step": 9783
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5127378616325606e-06,
+      "loss": 0.8469,
+      "step": 9784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5112865638353446e-06,
+      "loss": 0.8232,
+      "step": 9785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.509835905640431e-06,
+      "loss": 0.8456,
+      "step": 9786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5083858871571299e-06,
+      "loss": 0.8636,
+      "step": 9787
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5069365084946941e-06,
+      "loss": 0.842,
+      "step": 9788
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5054877697623305e-06,
+      "loss": 0.8681,
+      "step": 9789
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5040396710691996e-06,
+      "loss": 0.9008,
+      "step": 9790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5025922125244052e-06,
+      "loss": 0.9053,
+      "step": 9791
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5011453942370191e-06,
+      "loss": 0.8017,
+      "step": 9792
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4996992163160461e-06,
+      "loss": 0.8541,
+      "step": 9793
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4982536788704604e-06,
+      "loss": 0.839,
+      "step": 9794
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4968087820091714e-06,
+      "loss": 0.8786,
+      "step": 9795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4953645258410566e-06,
+      "loss": 0.874,
+      "step": 9796
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4939209104749274e-06,
+      "loss": 0.8483,
+      "step": 9797
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4924779360195662e-06,
+      "loss": 0.8761,
+      "step": 9798
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.491035602583688e-06,
+      "loss": 0.8239,
+      "step": 9799
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4895939102759782e-06,
+      "loss": 0.9013,
+      "step": 9800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4881528592050576e-06,
+      "loss": 0.8312,
+      "step": 9801
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4867124494795082e-06,
+      "loss": 0.8405,
+      "step": 9802
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4852726812078611e-06,
+      "loss": 0.841,
+      "step": 9803
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4838335544985982e-06,
+      "loss": 0.8432,
+      "step": 9804
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.482395069460154e-06,
+      "loss": 0.8498,
+      "step": 9805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4809572262009164e-06,
+      "loss": 0.8583,
+      "step": 9806
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4795200248292207e-06,
+      "loss": 0.8727,
+      "step": 9807
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4780834654533572e-06,
+      "loss": 0.8521,
+      "step": 9808
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.476647548181568e-06,
+      "loss": 0.8676,
+      "step": 9809
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.475212273122043e-06,
+      "loss": 0.8743,
+      "step": 9810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4737776403829296e-06,
+      "loss": 0.825,
+      "step": 9811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.472343650072322e-06,
+      "loss": 0.8719,
+      "step": 9812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4709103022982673e-06,
+      "loss": 0.8413,
+      "step": 9813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4694775971687648e-06,
+      "loss": 0.8902,
+      "step": 9814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4680455347917655e-06,
+      "loss": 0.8195,
+      "step": 9815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.466614115275171e-06,
+      "loss": 0.8802,
+      "step": 9816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4651833387268365e-06,
+      "loss": 0.8797,
+      "step": 9817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4637532052545644e-06,
+      "loss": 0.8249,
+      "step": 9818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.462323714966114e-06,
+      "loss": 0.8406,
+      "step": 9819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4608948679691926e-06,
+      "loss": 0.907,
+      "step": 9820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4594666643714629e-06,
+      "loss": 0.8546,
+      "step": 9821
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4580391042805287e-06,
+      "loss": 0.8216,
+      "step": 9822
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.456612187803963e-06,
+      "loss": 0.8446,
+      "step": 9823
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4551859150492697e-06,
+      "loss": 0.8678,
+      "step": 9824
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4537602861239253e-06,
+      "loss": 0.8943,
+      "step": 9825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.452335301135337e-06,
+      "loss": 0.8013,
+      "step": 9826
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4509109601908845e-06,
+      "loss": 0.8051,
+      "step": 9827
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4494872633978763e-06,
+      "loss": 0.8216,
+      "step": 9828
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4480642108635955e-06,
+      "loss": 0.8827,
+      "step": 9829
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4466418026952555e-06,
+      "loss": 0.868,
+      "step": 9830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.445220039000037e-06,
+      "loss": 0.8875,
+      "step": 9831
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4437989198850688e-06,
+      "loss": 0.5627,
+      "step": 9832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4423784454574185e-06,
+      "loss": 0.8585,
+      "step": 9833
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4409586158241272e-06,
+      "loss": 0.809,
+      "step": 9834
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4395394310921639e-06,
+      "loss": 0.8661,
+      "step": 9835
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4381208913684707e-06,
+      "loss": 0.8569,
+      "step": 9836
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4367029967599211e-06,
+      "loss": 0.5445,
+      "step": 9837
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4352857473733595e-06,
+      "loss": 0.8402,
+      "step": 9838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4338691433155627e-06,
+      "loss": 0.8194,
+      "step": 9839
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4324531846932766e-06,
+      "loss": 0.8142,
+      "step": 9840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4310378716131811e-06,
+      "loss": 0.4441,
+      "step": 9841
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4296232041819259e-06,
+      "loss": 0.841,
+      "step": 9842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4282091825060963e-06,
+      "loss": 0.8675,
+      "step": 9843
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.426795806692235e-06,
+      "loss": 0.8444,
+      "step": 9844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4253830768468379e-06,
+      "loss": 0.793,
+      "step": 9845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4239709930763513e-06,
+      "loss": 0.8724,
+      "step": 9846
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4225595554871707e-06,
+      "loss": 0.8772,
+      "step": 9847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4211487641856437e-06,
+      "loss": 0.5612,
+      "step": 9848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4197386192780715e-06,
+      "loss": 0.5838,
+      "step": 9849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4183291208707029e-06,
+      "loss": 0.8587,
+      "step": 9850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4169202690697414e-06,
+      "loss": 0.8549,
+      "step": 9851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4155120639813392e-06,
+      "loss": 0.8368,
+      "step": 9852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.414104505711602e-06,
+      "loss": 0.8449,
+      "step": 9853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4126975943665844e-06,
+      "loss": 0.8431,
+      "step": 9854
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4112913300522946e-06,
+      "loss": 0.8315,
+      "step": 9855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4098857128746912e-06,
+      "loss": 0.829,
+      "step": 9856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4084807429396818e-06,
+      "loss": 0.8809,
+      "step": 9857
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4070764203531283e-06,
+      "loss": 0.8961,
+      "step": 9858
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.405672745220844e-06,
+      "loss": 0.8887,
+      "step": 9859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4042697176485898e-06,
+      "loss": 0.8567,
+      "step": 9860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4028673377420821e-06,
+      "loss": 0.841,
+      "step": 9861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4014656056069854e-06,
+      "loss": 0.876,
+      "step": 9862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4000645213489194e-06,
+      "loss": 0.9156,
+      "step": 9863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3986640850734444e-06,
+      "loss": 0.533,
+      "step": 9864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3972642968860905e-06,
+      "loss": 0.826,
+      "step": 9865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3958651568923165e-06,
+      "loss": 0.8858,
+      "step": 9866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3944666651975559e-06,
+      "loss": 0.8002,
+      "step": 9867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3930688219071697e-06,
+      "loss": 0.8489,
+      "step": 9868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.391671627126493e-06,
+      "loss": 0.4621,
+      "step": 9869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.390275080960789e-06,
+      "loss": 0.869,
+      "step": 9870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.388879183515296e-06,
+      "loss": 0.8612,
+      "step": 9871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3874839348951796e-06,
+      "loss": 0.8909,
+      "step": 9872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3860893352055782e-06,
+      "loss": 0.8173,
+      "step": 9873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3846953845515655e-06,
+      "loss": 0.8672,
+      "step": 9874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3833020830381694e-06,
+      "loss": 0.8715,
+      "step": 9875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3819094307703807e-06,
+      "loss": 0.8602,
+      "step": 9876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.380517427853122e-06,
+      "loss": 0.8276,
+      "step": 9877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.379126074391287e-06,
+      "loss": 0.7976,
+      "step": 9878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3777353704897002e-06,
+      "loss": 0.8204,
+      "step": 9879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.376345316253157e-06,
+      "loss": 0.8494,
+      "step": 9880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3749559117863865e-06,
+      "loss": 0.84,
+      "step": 9881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3735671571940835e-06,
+      "loss": 0.7745,
+      "step": 9882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3721790525808797e-06,
+      "loss": 0.8363,
+      "step": 9883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.370791598051373e-06,
+      "loss": 0.8447,
+      "step": 9884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3694047937100985e-06,
+      "loss": 0.8841,
+      "step": 9885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3680186396615514e-06,
+      "loss": 0.792,
+      "step": 9886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3666331360101725e-06,
+      "loss": 0.8835,
+      "step": 9887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3652482828603575e-06,
+      "loss": 0.8961,
+      "step": 9888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3638640803164516e-06,
+      "loss": 0.9043,
+      "step": 9889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3624805284827492e-06,
+      "loss": 0.8679,
+      "step": 9890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3610976274634991e-06,
+      "loss": 0.843,
+      "step": 9891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.359715377362899e-06,
+      "loss": 0.8589,
+      "step": 9892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.358333778285097e-06,
+      "loss": 0.8751,
+      "step": 9893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3569528303341927e-06,
+      "loss": 0.8626,
+      "step": 9894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3555725336142378e-06,
+      "loss": 0.867,
+      "step": 9895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3541928882292344e-06,
+      "loss": 0.7717,
+      "step": 9896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3528138942831337e-06,
+      "loss": 0.8396,
+      "step": 9897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3514355518798417e-06,
+      "loss": 0.8551,
+      "step": 9898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3500578611232095e-06,
+      "loss": 0.8422,
+      "step": 9899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3486808221170455e-06,
+      "loss": 0.8419,
+      "step": 9900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3473044349651042e-06,
+      "loss": 0.8443,
+      "step": 9901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.345928699771093e-06,
+      "loss": 0.8301,
+      "step": 9902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3445536166386708e-06,
+      "loss": 0.8881,
+      "step": 9903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3431791856714471e-06,
+      "loss": 0.8695,
+      "step": 9904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3418054069729814e-06,
+      "loss": 0.8605,
+      "step": 9905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3404322806467796e-06,
+      "loss": 0.8921,
+      "step": 9906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3390598067963112e-06,
+      "loss": 0.853,
+      "step": 9907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3376879855249803e-06,
+      "loss": 0.8396,
+      "step": 9908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3363168169361574e-06,
+      "loss": 0.8548,
+      "step": 9909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3349463011331498e-06,
+      "loss": 0.8511,
+      "step": 9910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3335764382192307e-06,
+      "loss": 0.8813,
+      "step": 9911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3322072282976051e-06,
+      "loss": 0.8832,
+      "step": 9912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3308386714714495e-06,
+      "loss": 0.8379,
+      "step": 9913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3294707678438723e-06,
+      "loss": 0.8715,
+      "step": 9914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3281035175179503e-06,
+      "loss": 0.8546,
+      "step": 9915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3267369205966951e-06,
+      "loss": 0.859,
+      "step": 9916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.325370977183078e-06,
+      "loss": 0.8428,
+      "step": 9917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.324005687380021e-06,
+      "loss": 0.8491,
+      "step": 9918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3226410512903908e-06,
+      "loss": 0.8301,
+      "step": 9919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3212770690170174e-06,
+      "loss": 0.8621,
+      "step": 9920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3199137406626639e-06,
+      "loss": 0.8162,
+      "step": 9921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3185510663300615e-06,
+      "loss": 0.8426,
+      "step": 9922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3171890461218772e-06,
+      "loss": 0.8785,
+      "step": 9923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3158276801407432e-06,
+      "loss": 0.8441,
+      "step": 9924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3144669684892254e-06,
+      "loss": 0.8458,
+      "step": 9925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3131069112698602e-06,
+      "loss": 0.8363,
+      "step": 9926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3117475085851173e-06,
+      "loss": 0.8697,
+      "step": 9927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3103887605374255e-06,
+      "loss": 0.8309,
+      "step": 9928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3090306672291631e-06,
+      "loss": 0.4303,
+      "step": 9929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3076732287626603e-06,
+      "loss": 0.869,
+      "step": 9930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3063164452401943e-06,
+      "loss": 0.8584,
+      "step": 9931
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3049603167639969e-06,
+      "loss": 0.8028,
+      "step": 9932
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303604843436248e-06,
+      "loss": 0.8318,
+      "step": 9933
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3022500253590798e-06,
+      "loss": 0.8687,
+      "step": 9934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3008958626345735e-06,
+      "loss": 0.8626,
+      "step": 9935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2995423553647623e-06,
+      "loss": 0.8493,
+      "step": 9936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2981895036516278e-06,
+      "loss": 0.8505,
+      "step": 9937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2968373075971063e-06,
+      "loss": 0.8829,
+      "step": 9938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2954857673030807e-06,
+      "loss": 0.8195,
+      "step": 9939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2941348828713861e-06,
+      "loss": 0.8748,
+      "step": 9940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2927846544038091e-06,
+      "loss": 0.8378,
+      "step": 9941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2914350820020837e-06,
+      "loss": 0.8307,
+      "step": 9942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2900861657678998e-06,
+      "loss": 0.8803,
+      "step": 9943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2887379058028915e-06,
+      "loss": 0.8274,
+      "step": 9944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2873903022086487e-06,
+      "loss": 0.8632,
+      "step": 9945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2860433550867092e-06,
+      "loss": 0.8871,
+      "step": 9946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2846970645385638e-06,
+      "loss": 0.8637,
+      "step": 9947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2833514306656468e-06,
+      "loss": 0.5121,
+      "step": 9948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2820064535693543e-06,
+      "loss": 0.8453,
+      "step": 9949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2806621333510204e-06,
+      "loss": 0.8351,
+      "step": 9950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2793184701119444e-06,
+      "loss": 0.8775,
+      "step": 9951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2779754639533592e-06,
+      "loss": 0.8362,
+      "step": 9952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2766331149764643e-06,
+      "loss": 0.9059,
+      "step": 9953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2752914232823942e-06,
+      "loss": 0.8143,
+      "step": 9954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2739503889722505e-06,
+      "loss": 0.8725,
+      "step": 9955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2726100121470676e-06,
+      "loss": 0.8604,
+      "step": 9956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.271270292907849e-06,
+      "loss": 0.8581,
+      "step": 9957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2699312313555323e-06,
+      "loss": 0.8554,
+      "step": 9958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2685928275910142e-06,
+      "loss": 0.8417,
+      "step": 9959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2672550817151397e-06,
+      "loss": 0.828,
+      "step": 9960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2659179938287035e-06,
+      "loss": 0.8632,
+      "step": 9961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2645815640324545e-06,
+      "loss": 0.8635,
+      "step": 9962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2632457924270835e-06,
+      "loss": 0.8207,
+      "step": 9963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2619106791132475e-06,
+      "loss": 0.878,
+      "step": 9964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.260576224191532e-06,
+      "loss": 0.7981,
+      "step": 9965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2592424277624948e-06,
+      "loss": 0.8692,
+      "step": 9966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2579092899266254e-06,
+      "loss": 0.856,
+      "step": 9967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2565768107843812e-06,
+      "loss": 0.8558,
+      "step": 9968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.255244990436153e-06,
+      "loss": 0.8483,
+      "step": 9969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2539138289822939e-06,
+      "loss": 0.8938,
+      "step": 9970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2525833265231025e-06,
+      "loss": 0.8787,
+      "step": 9971
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2512534831588285e-06,
+      "loss": 0.8243,
+      "step": 9972
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.249924298989672e-06,
+      "loss": 0.901,
+      "step": 9973
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.248595774115784e-06,
+      "loss": 0.8672,
+      "step": 9974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2472679086372662e-06,
+      "loss": 0.8058,
+      "step": 9975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.245940702654168e-06,
+      "loss": 0.8267,
+      "step": 9976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2446141562664927e-06,
+      "loss": 0.842,
+      "step": 9977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.243288269574191e-06,
+      "loss": 0.8163,
+      "step": 9978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2419630426771644e-06,
+      "loss": 0.8519,
+      "step": 9979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2406384756752665e-06,
+      "loss": 0.8005,
+      "step": 9980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2393145686682995e-06,
+      "loss": 0.8586,
+      "step": 9981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2379913217560157e-06,
+      "loss": 0.8496,
+      "step": 9982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2366687350381202e-06,
+      "loss": 0.8351,
+      "step": 9983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2353468086142639e-06,
+      "loss": 0.8208,
+      "step": 9984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.234025542584053e-06,
+      "loss": 0.8557,
+      "step": 9985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.232704937047039e-06,
+      "loss": 0.8519,
+      "step": 9986
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2313849921027277e-06,
+      "loss": 0.8297,
+      "step": 9987
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2300657078505728e-06,
+      "loss": 0.8185,
+      "step": 9988
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.228747084389982e-06,
+      "loss": 0.8172,
+      "step": 9989
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2274291218203027e-06,
+      "loss": 0.8096,
+      "step": 9990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2261118202408484e-06,
+      "loss": 0.933,
+      "step": 9991
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2247951797508662e-06,
+      "loss": 0.9299,
+      "step": 9992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2234792004495699e-06,
+      "loss": 0.8666,
+      "step": 9993
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.222163882436107e-06,
+      "loss": 0.812,
+      "step": 9994
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2208492258095917e-06,
+      "loss": 0.8059,
+      "step": 9995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2195352306690711e-06,
+      "loss": 0.8653,
+      "step": 9996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2182218971135595e-06,
+      "loss": 0.8511,
+      "step": 9997
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2169092252420046e-06,
+      "loss": 0.8658,
+      "step": 9998
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2155972151533225e-06,
+      "loss": 0.8211,
+      "step": 9999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.214285866946362e-06,
+      "loss": 0.8484,
+      "step": 10000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2129751807199319e-06,
+      "loss": 0.8209,
+      "step": 10001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.21166515657279e-06,
+      "loss": 0.8892,
+      "step": 10002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2103557946036425e-06,
+      "loss": 0.8184,
+      "step": 10003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2090470949111465e-06,
+      "loss": 0.877,
+      "step": 10004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2077390575939097e-06,
+      "loss": 0.8527,
+      "step": 10005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2064316827504885e-06,
+      "loss": 0.4718,
+      "step": 10006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2051249704793877e-06,
+      "loss": 0.8516,
+      "step": 10007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2038189208790718e-06,
+      "loss": 0.8526,
+      "step": 10008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2025135340479398e-06,
+      "loss": 0.8552,
+      "step": 10009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2012088100843577e-06,
+      "loss": 0.831,
+      "step": 10010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1999047490866255e-06,
+      "loss": 0.8422,
+      "step": 10011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1986013511530048e-06,
+      "loss": 0.8865,
+      "step": 10012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1972986163817012e-06,
+      "loss": 0.8824,
+      "step": 10013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1959965448708731e-06,
+      "loss": 0.8386,
+      "step": 10014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1946951367186288e-06,
+      "loss": 0.8657,
+      "step": 10015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1933943920230262e-06,
+      "loss": 0.899,
+      "step": 10016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1920943108820714e-06,
+      "loss": 0.8516,
+      "step": 10017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.190794893393723e-06,
+      "loss": 0.8561,
+      "step": 10018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.189496139655889e-06,
+      "loss": 0.8515,
+      "step": 10019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1881980497664282e-06,
+      "loss": 0.8693,
+      "step": 10020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.186900623823145e-06,
+      "loss": 0.872,
+      "step": 10021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1856038619237998e-06,
+      "loss": 0.5437,
+      "step": 10022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1843077641660994e-06,
+      "loss": 0.8734,
+      "step": 10023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1830123306477015e-06,
+      "loss": 0.8507,
+      "step": 10024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1817175614662136e-06,
+      "loss": 0.8862,
+      "step": 10025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1804234567191919e-06,
+      "loss": 0.8221,
+      "step": 10026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1791300165041463e-06,
+      "loss": 0.8656,
+      "step": 10027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1778372409185324e-06,
+      "loss": 0.883,
+      "step": 10028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1765451300597574e-06,
+      "loss": 0.8486,
+      "step": 10029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1752536840251794e-06,
+      "loss": 0.8608,
+      "step": 10030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1739629029121058e-06,
+      "loss": 0.823,
+      "step": 10031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1726727868177902e-06,
+      "loss": 0.8275,
+      "step": 10032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1713833358394456e-06,
+      "loss": 0.8976,
+      "step": 10033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1700945500742212e-06,
+      "loss": 0.8248,
+      "step": 10034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1688064296192313e-06,
+      "loss": 0.834,
+      "step": 10035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1675189745715243e-06,
+      "loss": 0.8127,
+      "step": 10036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1662321850281155e-06,
+      "loss": 0.8515,
+      "step": 10037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.164946061085952e-06,
+      "loss": 0.8064,
+      "step": 10038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1636606028419495e-06,
+      "loss": 0.8622,
+      "step": 10039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1623758103929538e-06,
+      "loss": 0.8694,
+      "step": 10040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.16109168383578e-06,
+      "loss": 0.8786,
+      "step": 10041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.159808223267178e-06,
+      "loss": 0.8747,
+      "step": 10042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1585254287838544e-06,
+      "loss": 0.8455,
+      "step": 10043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1572433004824635e-06,
+      "loss": 0.7586,
+      "step": 10044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1559618384596127e-06,
+      "loss": 0.9445,
+      "step": 10045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1546810428118548e-06,
+      "loss": 0.8114,
+      "step": 10046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.153400913635695e-06,
+      "loss": 0.8042,
+      "step": 10047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.152121451027587e-06,
+      "loss": 0.8298,
+      "step": 10048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1508426550839368e-06,
+      "loss": 0.8161,
+      "step": 10049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1495645259010969e-06,
+      "loss": 0.8759,
+      "step": 10050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1482870635753696e-06,
+      "loss": 0.8895,
+      "step": 10051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1470102682030139e-06,
+      "loss": 0.8292,
+      "step": 10052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1457341398802269e-06,
+      "loss": 0.8739,
+      "step": 10053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1444586787031632e-06,
+      "loss": 0.8242,
+      "step": 10054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1431838847679265e-06,
+      "loss": 0.8238,
+      "step": 10055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1419097581705686e-06,
+      "loss": 0.8671,
+      "step": 10056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1406362990070918e-06,
+      "loss": 0.8483,
+      "step": 10057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1393635073734476e-06,
+      "loss": 0.8508,
+      "step": 10058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1380913833655383e-06,
+      "loss": 0.827,
+      "step": 10059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1368199270792135e-06,
+      "loss": 0.8653,
+      "step": 10060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1355491386102757e-06,
+      "loss": 0.8509,
+      "step": 10061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.134279018054475e-06,
+      "loss": 0.8558,
+      "step": 10062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1330095655075113e-06,
+      "loss": 0.8823,
+      "step": 10063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1317407810650372e-06,
+      "loss": 0.8488,
+      "step": 10064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.130472664822646e-06,
+      "loss": 0.8187,
+      "step": 10065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1292052168758916e-06,
+      "loss": 0.8347,
+      "step": 10066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.127938437320273e-06,
+      "loss": 0.8442,
+      "step": 10067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.126672326251238e-06,
+      "loss": 0.8508,
+      "step": 10068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.125406883764184e-06,
+      "loss": 0.7966,
+      "step": 10069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.124142109954459e-06,
+      "loss": 0.5236,
+      "step": 10070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1228780049173616e-06,
+      "loss": 0.8593,
+      "step": 10071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1216145687481373e-06,
+      "loss": 0.8858,
+      "step": 10072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1203518015419855e-06,
+      "loss": 0.7903,
+      "step": 10073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1190897033940461e-06,
+      "loss": 0.8414,
+      "step": 10074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1178282743994228e-06,
+      "loss": 0.8668,
+      "step": 10075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.116567514653153e-06,
+      "loss": 0.86,
+      "step": 10076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1153074242502404e-06,
+      "loss": 0.8608,
+      "step": 10077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1140480032856204e-06,
+      "loss": 0.8171,
+      "step": 10078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1127892518541961e-06,
+      "loss": 0.882,
+      "step": 10079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1115311700508026e-06,
+      "loss": 0.7753,
+      "step": 10080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1102737579702404e-06,
+      "loss": 0.8326,
+      "step": 10081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1090170157072455e-06,
+      "loss": 0.8971,
+      "step": 10082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1077609433565173e-06,
+      "loss": 0.854,
+      "step": 10083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1065055410126913e-06,
+      "loss": 0.8181,
+      "step": 10084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1052508087703617e-06,
+      "loss": 0.8631,
+      "step": 10085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1039967467240687e-06,
+      "loss": 0.875,
+      "step": 10086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.102743354968303e-06,
+      "loss": 0.8536,
+      "step": 10087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.101490633597504e-06,
+      "loss": 0.8634,
+      "step": 10088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1002385827060602e-06,
+      "loss": 0.8376,
+      "step": 10089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0989872023883118e-06,
+      "loss": 0.8638,
+      "step": 10090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.097736492738547e-06,
+      "loss": 0.8867,
+      "step": 10091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0964864538510022e-06,
+      "loss": 0.8622,
+      "step": 10092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0952370858198657e-06,
+      "loss": 0.8694,
+      "step": 10093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0939883887392743e-06,
+      "loss": 0.837,
+      "step": 10094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0927403627033129e-06,
+      "loss": 0.8465,
+      "step": 10095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.091493007806017e-06,
+      "loss": 0.8535,
+      "step": 10096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0902463241413718e-06,
+      "loss": 0.8437,
+      "step": 10097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0890003118033132e-06,
+      "loss": 0.8727,
+      "step": 10098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0877549708857228e-06,
+      "loss": 0.8659,
+      "step": 10099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0865103014824351e-06,
+      "loss": 0.871,
+      "step": 10100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0852663036872324e-06,
+      "loss": 0.8652,
+      "step": 10101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.084022977593846e-06,
+      "loss": 0.8303,
+      "step": 10102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0827803232959577e-06,
+      "loss": 0.8738,
+      "step": 10103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0815383408871983e-06,
+      "loss": 0.8245,
+      "step": 10104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0802970304611481e-06,
+      "loss": 0.869,
+      "step": 10105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0790563921113384e-06,
+      "loss": 0.867,
+      "step": 10106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0778164259312418e-06,
+      "loss": 0.8509,
+      "step": 10107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0765771320142958e-06,
+      "loss": 0.9003,
+      "step": 10108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0753385104538682e-06,
+      "loss": 0.8288,
+      "step": 10109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.074100561343292e-06,
+      "loss": 0.8576,
+      "step": 10110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0728632847758425e-06,
+      "loss": 0.8019,
+      "step": 10111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0716266808447452e-06,
+      "loss": 0.8296,
+      "step": 10112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0703907496431743e-06,
+      "loss": 0.9039,
+      "step": 10113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0691554912642532e-06,
+      "loss": 0.8481,
+      "step": 10114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0679209058010597e-06,
+      "loss": 0.8585,
+      "step": 10115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0666869933466085e-06,
+      "loss": 0.825,
+      "step": 10116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0654537539938814e-06,
+      "loss": 0.8836,
+      "step": 10117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06422118783579e-06,
+      "loss": 0.8634,
+      "step": 10118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0629892949652133e-06,
+      "loss": 0.8504,
+      "step": 10119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0617580754749646e-06,
+      "loss": 0.873,
+      "step": 10120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06052752945782e-06,
+      "loss": 0.4445,
+      "step": 10121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0592976570064894e-06,
+      "loss": 0.844,
+      "step": 10122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.058068458213649e-06,
+      "loss": 0.8733,
+      "step": 10123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0568399331719082e-06,
+      "loss": 0.8845,
+      "step": 10124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0556120819738403e-06,
+      "loss": 0.8144,
+      "step": 10125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0543849047119548e-06,
+      "loss": 0.8288,
+      "step": 10126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0531584014787198e-06,
+      "loss": 0.8562,
+      "step": 10127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0519325723665463e-06,
+      "loss": 0.8609,
+      "step": 10128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0507074174678e-06,
+      "loss": 0.8489,
+      "step": 10129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0494829368747917e-06,
+      "loss": 0.8779,
+      "step": 10130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0482591306797829e-06,
+      "loss": 0.8952,
+      "step": 10131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0470359989749845e-06,
+      "loss": 0.8566,
+      "step": 10132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0458135418525572e-06,
+      "loss": 0.8434,
+      "step": 10133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0445917594046073e-06,
+      "loss": 0.8482,
+      "step": 10134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0433706517231957e-06,
+      "loss": 0.8827,
+      "step": 10135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0421502189003274e-06,
+      "loss": 0.8156,
+      "step": 10136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0409304610279603e-06,
+      "loss": 0.8829,
+      "step": 10137
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0397113781979994e-06,
+      "loss": 0.8615,
+      "step": 10138
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0384929705022994e-06,
+      "loss": 0.8543,
+      "step": 10139
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0372752380326645e-06,
+      "loss": 0.8699,
+      "step": 10140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0360581808808478e-06,
+      "loss": 0.9022,
+      "step": 10141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0348417991385506e-06,
+      "loss": 0.8512,
+      "step": 10142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0336260928974252e-06,
+      "loss": 0.848,
+      "step": 10143
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0324110622490702e-06,
+      "loss": 0.8307,
+      "step": 10144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0311967072850359e-06,
+      "loss": 0.8401,
+      "step": 10145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0299830280968205e-06,
+      "loss": 0.8295,
+      "step": 10146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0287700247758725e-06,
+      "loss": 0.8296,
+      "step": 10147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0275576974135893e-06,
+      "loss": 0.8072,
+      "step": 10148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.026346046101312e-06,
+      "loss": 0.8562,
+      "step": 10149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0251350709303409e-06,
+      "loss": 0.8477,
+      "step": 10150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0239247719919144e-06,
+      "loss": 0.8905,
+      "step": 10151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0227151493772324e-06,
+      "loss": 0.8833,
+      "step": 10152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0215062031774293e-06,
+      "loss": 0.8783,
+      "step": 10153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0202979334836006e-06,
+      "loss": 0.8792,
+      "step": 10154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0190903403867847e-06,
+      "loss": 0.5536,
+      "step": 10155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.017883423977971e-06,
+      "loss": 0.7842,
+      "step": 10156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0166771843481004e-06,
+      "loss": 0.857,
+      "step": 10157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0154716215880523e-06,
+      "loss": 0.8451,
+      "step": 10158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0142667357886714e-06,
+      "loss": 0.8434,
+      "step": 10159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0130625270407347e-06,
+      "loss": 0.8606,
+      "step": 10160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0118589954349845e-06,
+      "loss": 0.8968,
+      "step": 10161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0106561410620953e-06,
+      "loss": 0.85,
+      "step": 10162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0094539640127067e-06,
+      "loss": 0.8584,
+      "step": 10163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0082524643773916e-06,
+      "loss": 0.8695,
+      "step": 10164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0070516422466892e-06,
+      "loss": 0.8966,
+      "step": 10165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.005851497711069e-06,
+      "loss": 0.8449,
+      "step": 10166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0046520308609664e-06,
+      "loss": 0.8924,
+      "step": 10167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0034532417867537e-06,
+      "loss": 0.838,
+      "step": 10168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0022551305787564e-06,
+      "loss": 0.8744,
+      "step": 10169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0010576973272512e-06,
+      "loss": 0.8886,
+      "step": 10170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.998609421224604e-07,
+      "loss": 0.8711,
+      "step": 10171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.986648650545561e-07,
+      "loss": 0.88,
+      "step": 10172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.974694662136609e-07,
+      "loss": 0.8665,
+      "step": 10173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.962747456898426e-07,
+      "loss": 0.8595,
+      "step": 10174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.950807035731224e-07,
+      "loss": 0.8638,
+      "step": 10175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.938873399534688e-07,
+      "loss": 0.8962,
+      "step": 10176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.92694654920796e-07,
+      "loss": 0.8703,
+      "step": 10177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.915026485649714e-07,
+      "loss": 0.8587,
+      "step": 10178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.8612,
+      "step": 10179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.891206722430735e-07,
+      "loss": 0.8775,
+      "step": 10180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.879307024564744e-07,
+      "loss": 0.8699,
+      "step": 10181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.867414117056763e-07,
+      "loss": 0.8532,
+      "step": 10182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.85552800080286e-07,
+      "loss": 0.7986,
+      "step": 10183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.843648676698648e-07,
+      "loss": 0.849,
+      "step": 10184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.831776145639182e-07,
+      "loss": 0.8183,
+      "step": 10185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.81991040851904e-07,
+      "loss": 0.8473,
+      "step": 10186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.808051466232261e-07,
+      "loss": 0.8327,
+      "step": 10187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.796199319672416e-07,
+      "loss": 0.8686,
+      "step": 10188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.784353969732486e-07,
+      "loss": 0.8081,
+      "step": 10189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.77251541730504e-07,
+      "loss": 0.7935,
+      "step": 10190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.760683663282012e-07,
+      "loss": 0.8943,
+      "step": 10191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.748858708554975e-07,
+      "loss": 0.8586,
+      "step": 10192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.737040554014831e-07,
+      "loss": 0.8499,
+      "step": 10193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.725229200552123e-07,
+      "loss": 0.8545,
+      "step": 10194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.713424649056735e-07,
+      "loss": 0.8744,
+      "step": 10195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.70162690041817e-07,
+      "loss": 0.849,
+      "step": 10196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.689835955525307e-07,
+      "loss": 0.8578,
+      "step": 10197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.678051815266587e-07,
+      "loss": 0.8392,
+      "step": 10198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.666274480529936e-07,
+      "loss": 0.8724,
+      "step": 10199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.654503952202687e-07,
+      "loss": 0.8528,
+      "step": 10200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.642740231171787e-07,
+      "loss": 0.827,
+      "step": 10201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.630983318323527e-07,
+      "loss": 0.8415,
+      "step": 10202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.619233214543833e-07,
+      "loss": 0.8224,
+      "step": 10203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.607489920717983e-07,
+      "loss": 0.8358,
+      "step": 10204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59575343773087e-07,
+      "loss": 0.8499,
+      "step": 10205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.584023766466721e-07,
+      "loss": 0.8535,
+      "step": 10206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.572300907809428e-07,
+      "loss": 0.8667,
+      "step": 10207
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.560584862642197e-07,
+      "loss": 0.8842,
+      "step": 10208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.548875631847875e-07,
+      "loss": 0.8136,
+      "step": 10209
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.53717321630866e-07,
+      "loss": 0.8457,
+      "step": 10210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.525477616906331e-07,
+      "loss": 0.8416,
+      "step": 10211
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.513788834522108e-07,
+      "loss": 0.8361,
+      "step": 10212
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.502106870036709e-07,
+      "loss": 0.7848,
+      "step": 10213
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.490431724330352e-07,
+      "loss": 0.886,
+      "step": 10214
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.47876339828272e-07,
+      "loss": 0.8531,
+      "step": 10215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.46710189277299e-07,
+      "loss": 0.8672,
+      "step": 10216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.455447208679836e-07,
+      "loss": 0.8448,
+      "step": 10217
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.443799346881388e-07,
+      "loss": 0.8557,
+      "step": 10218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.432158308255301e-07,
+      "loss": 0.8782,
+      "step": 10219
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.420524093678696e-07,
+      "loss": 0.832,
+      "step": 10220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.40889670402817e-07,
+      "loss": 0.8329,
+      "step": 10221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.397276140179823e-07,
+      "loss": 0.5177,
+      "step": 10222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.385662403009222e-07,
+      "loss": 0.872,
+      "step": 10223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.374055493391455e-07,
+      "loss": 0.8362,
+      "step": 10224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.362455412201043e-07,
+      "loss": 0.8377,
+      "step": 10225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.350862160312046e-07,
+      "loss": 0.8503,
+      "step": 10226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.339275738597975e-07,
+      "loss": 0.8338,
+      "step": 10227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.327696147931842e-07,
+      "loss": 0.909,
+      "step": 10228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.316123389186138e-07,
+      "loss": 0.8466,
+      "step": 10229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.304557463232844e-07,
+      "loss": 0.8733,
+      "step": 10230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.292998370943385e-07,
+      "loss": 0.8685,
+      "step": 10231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.281446113188774e-07,
+      "loss": 0.8546,
+      "step": 10232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.269900690839373e-07,
+      "loss": 0.8327,
+      "step": 10233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.258362104765173e-07,
+      "loss": 0.849,
+      "step": 10234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.24683035583549e-07,
+      "loss": 0.8573,
+      "step": 10235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.235305444919307e-07,
+      "loss": 0.8316,
+      "step": 10236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.223787372884896e-07,
+      "loss": 0.8609,
+      "step": 10237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.212276140600218e-07,
+      "loss": 0.8946,
+      "step": 10238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.200771748932513e-07,
+      "loss": 0.8605,
+      "step": 10239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.189274198748688e-07,
+      "loss": 0.8461,
+      "step": 10240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.177783490915015e-07,
+      "loss": 0.861,
+      "step": 10241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.166299626297271e-07,
+      "loss": 0.8233,
+      "step": 10242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.154822605760782e-07,
+      "loss": 0.8721,
+      "step": 10243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.14335243017026e-07,
+      "loss": 0.9095,
+      "step": 10244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.131889100390024e-07,
+      "loss": 0.891,
+      "step": 10245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.120432617283714e-07,
+      "loss": 0.8441,
+      "step": 10246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.108982981714631e-07,
+      "loss": 0.8751,
+      "step": 10247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.097540194545407e-07,
+      "loss": 0.8354,
+      "step": 10248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.086104256638284e-07,
+      "loss": 0.8043,
+      "step": 10249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.074675168854874e-07,
+      "loss": 0.8098,
+      "step": 10250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.063252932056399e-07,
+      "loss": 0.8771,
+      "step": 10251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.051837547103426e-07,
+      "loss": 0.8471,
+      "step": 10252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.0404290148561e-07,
+      "loss": 0.8806,
+      "step": 10253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.029027336174023e-07,
+      "loss": 0.848,
+      "step": 10254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.017632511916296e-07,
+      "loss": 0.5077,
+      "step": 10255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.006244542941467e-07,
+      "loss": 0.9047,
+      "step": 10256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.994863430107603e-07,
+      "loss": 0.8065,
+      "step": 10257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.983489174272242e-07,
+      "loss": 0.8297,
+      "step": 10258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.972121776292386e-07,
+      "loss": 0.9022,
+      "step": 10259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.960761237024573e-07,
+      "loss": 0.8261,
+      "step": 10260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.949407557324752e-07,
+      "loss": 0.8145,
+      "step": 10261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.938060738048415e-07,
+      "loss": 0.8433,
+      "step": 10262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.926720780050513e-07,
+      "loss": 0.8301,
+      "step": 10263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.915387684185484e-07,
+      "loss": 0.8526,
+      "step": 10264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.904061451307234e-07,
+      "loss": 0.8395,
+      "step": 10265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.892742082269179e-07,
+      "loss": 0.8809,
+      "step": 10266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.881429577924206e-07,
+      "loss": 0.8608,
+      "step": 10267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.870123939124665e-07,
+      "loss": 0.8301,
+      "step": 10268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.858825166722418e-07,
+      "loss": 0.8139,
+      "step": 10269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.847533261568797e-07,
+      "loss": 0.8334,
+      "step": 10270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.83624822451461e-07,
+      "loss": 0.8907,
+      "step": 10271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.824970056410187e-07,
+      "loss": 0.5182,
+      "step": 10272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.813698758105249e-07,
+      "loss": 0.8885,
+      "step": 10273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.802434330449128e-07,
+      "loss": 0.7927,
+      "step": 10274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.7911767742905e-07,
+      "loss": 0.8665,
+      "step": 10275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.779926090477653e-07,
+      "loss": 0.8108,
+      "step": 10276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.768682279858242e-07,
+      "loss": 0.847,
+      "step": 10277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.757445343279514e-07,
+      "loss": 0.9017,
+      "step": 10278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.746215281588088e-07,
+      "loss": 0.8234,
+      "step": 10279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.734992095630179e-07,
+      "loss": 0.8857,
+      "step": 10280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.723775786251354e-07,
+      "loss": 0.8174,
+      "step": 10281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.712566354296814e-07,
+      "loss": 0.8481,
+      "step": 10282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.701363800611096e-07,
+      "loss": 0.86,
+      "step": 10283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.690168126038301e-07,
+      "loss": 0.8544,
+      "step": 10284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.678979331421999e-07,
+      "loss": 0.8448,
+      "step": 10285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.667797417605228e-07,
+      "loss": 0.8255,
+      "step": 10286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.656622385430547e-07,
+      "loss": 0.8024,
+      "step": 10287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 0.8763,
+      "step": 10288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.634292969374869e-07,
+      "loss": 0.8631,
+      "step": 10289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.623138587176327e-07,
+      "loss": 0.8651,
+      "step": 10290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.611991089984816e-07,
+      "loss": 0.808,
+      "step": 10291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.600850478640199e-07,
+      "loss": 0.9349,
+      "step": 10292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.589716753981958e-07,
+      "loss": 0.896,
+      "step": 10293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.578589916848933e-07,
+      "loss": 0.8862,
+      "step": 10294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.567469968079534e-07,
+      "loss": 0.898,
+      "step": 10295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.556356908511598e-07,
+      "loss": 0.8531,
+      "step": 10296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.545250738982491e-07,
+      "loss": 0.8913,
+      "step": 10297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.53415146032901e-07,
+      "loss": 0.8535,
+      "step": 10298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.523059073387474e-07,
+      "loss": 0.82,
+      "step": 10299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.511973578993649e-07,
+      "loss": 0.4529,
+      "step": 10300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.500894977982821e-07,
+      "loss": 0.8324,
+      "step": 10301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.489823271189712e-07,
+      "loss": 0.855,
+      "step": 10302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.478758459448555e-07,
+      "loss": 0.855,
+      "step": 10303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.467700543593049e-07,
+      "loss": 0.8513,
+      "step": 10304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.456649524456384e-07,
+      "loss": 0.8641,
+      "step": 10305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.445605402871226e-07,
+      "loss": 0.8576,
+      "step": 10306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.434568179669722e-07,
+      "loss": 0.8045,
+      "step": 10307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.423537855683494e-07,
+      "loss": 0.809,
+      "step": 10308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.412514431743657e-07,
+      "loss": 0.8637,
+      "step": 10309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.401497908680778e-07,
+      "loss": 0.8609,
+      "step": 10310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.390488287324938e-07,
+      "loss": 0.8671,
+      "step": 10311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.379485568505686e-07,
+      "loss": 0.8164,
+      "step": 10312
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.368489753052034e-07,
+      "loss": 0.803,
+      "step": 10313
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.35750084179251e-07,
+      "loss": 0.8954,
+      "step": 10314
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.346518835555062e-07,
+      "loss": 0.5751,
+      "step": 10315
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.335543735167206e-07,
+      "loss": 0.8796,
+      "step": 10316
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.324575541455815e-07,
+      "loss": 0.8226,
+      "step": 10317
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.313614255247404e-07,
+      "loss": 0.8643,
+      "step": 10318
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.302659877367803e-07,
+      "loss": 0.8653,
+      "step": 10319
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.29171240864245e-07,
+      "loss": 0.4979,
+      "step": 10320
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.280771849896141e-07,
+      "loss": 0.5745,
+      "step": 10321
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.269838201953306e-07,
+      "loss": 0.862,
+      "step": 10322
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.258911465637675e-07,
+      "loss": 0.849,
+      "step": 10323
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.247991641772634e-07,
+      "loss": 0.8276,
+      "step": 10324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.237078731180914e-07,
+      "loss": 0.8423,
+      "step": 10325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.226172734684779e-07,
+      "loss": 0.8474,
+      "step": 10326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.215273653105959e-07,
+      "loss": 0.8699,
+      "step": 10327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.2043814872657e-07,
+      "loss": 0.8618,
+      "step": 10328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.193496237984677e-07,
+      "loss": 0.8585,
+      "step": 10329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.182617906083046e-07,
+      "loss": 0.8472,
+      "step": 10330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.171746492380528e-07,
+      "loss": 0.4839,
+      "step": 10331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.160881997696169e-07,
+      "loss": 0.8385,
+      "step": 10332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.150024422848668e-07,
+      "loss": 0.8415,
+      "step": 10333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.139173768656028e-07,
+      "loss": 0.9046,
+      "step": 10334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.128330035935906e-07,
+      "loss": 0.8581,
+      "step": 10335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.11749322550528e-07,
+      "loss": 0.89,
+      "step": 10336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.106663338180687e-07,
+      "loss": 0.8564,
+      "step": 10337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.095840374778153e-07,
+      "loss": 0.8623,
+      "step": 10338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.085024336113145e-07,
+      "loss": 0.8992,
+      "step": 10339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.074215223000625e-07,
+      "loss": 0.8305,
+      "step": 10340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.063413036255041e-07,
+      "loss": 0.8616,
+      "step": 10341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.052617776690285e-07,
+      "loss": 0.8646,
+      "step": 10342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.041829445119786e-07,
+      "loss": 0.5509,
+      "step": 10343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.031048042356393e-07,
+      "loss": 0.8357,
+      "step": 10344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.020273569212456e-07,
+      "loss": 0.8831,
+      "step": 10345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.009506026499813e-07,
+      "loss": 0.8674,
+      "step": 10346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.998745415029762e-07,
+      "loss": 0.866,
+      "step": 10347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.987991735613098e-07,
+      "loss": 0.8592,
+      "step": 10348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.977244989060073e-07,
+      "loss": 0.8581,
+      "step": 10349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.966505176180428e-07,
+      "loss": 0.8598,
+      "step": 10350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.955772297783382e-07,
+      "loss": 0.8434,
+      "step": 10351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.945046354677621e-07,
+      "loss": 0.8484,
+      "step": 10352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.934327347671333e-07,
+      "loss": 0.8682,
+      "step": 10353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.92361527757215e-07,
+      "loss": 0.887,
+      "step": 10354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.912910145187213e-07,
+      "loss": 0.8152,
+      "step": 10355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.902211951323135e-07,
+      "loss": 0.8536,
+      "step": 10356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.891520696785937e-07,
+      "loss": 0.8712,
+      "step": 10357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.880836382381262e-07,
+      "loss": 0.8371,
+      "step": 10358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.870159008914069e-07,
+      "loss": 0.8504,
+      "step": 10359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.859488577188945e-07,
+      "loss": 0.4709,
+      "step": 10360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.848825088009793e-07,
+      "loss": 0.8044,
+      "step": 10361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.838168542180169e-07,
+      "loss": 0.8196,
+      "step": 10362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.827518940502931e-07,
+      "loss": 0.8181,
+      "step": 10363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.816876283780583e-07,
+      "loss": 0.8512,
+      "step": 10364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.806240572814927e-07,
+      "loss": 0.8611,
+      "step": 10365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.795611808407433e-07,
+      "loss": 0.8391,
+      "step": 10366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.784989991358882e-07,
+      "loss": 0.8424,
+      "step": 10367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.774375122469624e-07,
+      "loss": 0.8602,
+      "step": 10368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.763767202539462e-07,
+      "loss": 0.7982,
+      "step": 10369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.753166232367671e-07,
+      "loss": 0.8409,
+      "step": 10370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.742572212753008e-07,
+      "loss": 0.8635,
+      "step": 10371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.731985144493703e-07,
+      "loss": 0.8571,
+      "step": 10372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.721405028387463e-07,
+      "loss": 0.8446,
+      "step": 10373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.710831865231461e-07,
+      "loss": 0.8513,
+      "step": 10374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.700265655822403e-07,
+      "loss": 0.8902,
+      "step": 10375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.689706400956354e-07,
+      "loss": 0.5545,
+      "step": 10376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.679154101428998e-07,
+      "loss": 0.8624,
+      "step": 10377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.668608758035379e-07,
+      "loss": 0.8287,
+      "step": 10378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.65807037157007e-07,
+      "loss": 0.8569,
+      "step": 10379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.647538942827115e-07,
+      "loss": 0.5171,
+      "step": 10380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.637014472600024e-07,
+      "loss": 0.8629,
+      "step": 10381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.626496961681795e-07,
+      "loss": 0.8509,
+      "step": 10382
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.615986410864895e-07,
+      "loss": 0.8971,
+      "step": 10383
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.605482820941257e-07,
+      "loss": 0.8141,
+      "step": 10384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.594986192702303e-07,
+      "loss": 0.8532,
+      "step": 10385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.584496526938933e-07,
+      "loss": 0.84,
+      "step": 10386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.574013824441506e-07,
+      "loss": 0.891,
+      "step": 10387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.563538085999878e-07,
+      "loss": 0.8748,
+      "step": 10388
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.55306931240335e-07,
+      "loss": 0.8357,
+      "step": 10389
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.542607504440736e-07,
+      "loss": 0.8497,
+      "step": 10390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.532152662900294e-07,
+      "loss": 0.8163,
+      "step": 10391
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.521704788569783e-07,
+      "loss": 0.8609,
+      "step": 10392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.511263882236408e-07,
+      "loss": 0.8316,
+      "step": 10393
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.50082994468686e-07,
+      "loss": 0.8134,
+      "step": 10394
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.490402976707323e-07,
+      "loss": 0.847,
+      "step": 10395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.479982979083444e-07,
+      "loss": 0.8897,
+      "step": 10396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.46956995260033e-07,
+      "loss": 0.8528,
+      "step": 10397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.459163898042599e-07,
+      "loss": 0.8994,
+      "step": 10398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.448764816194254e-07,
+      "loss": 0.8438,
+      "step": 10399
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.438372707838937e-07,
+      "loss": 0.8715,
+      "step": 10400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.427987573759576e-07,
+      "loss": 0.7666,
+      "step": 10401
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.417609414738725e-07,
+      "loss": 0.8937,
+      "step": 10402
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.407238231558312e-07,
+      "loss": 0.8132,
+      "step": 10403
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.396874024999811e-07,
+      "loss": 0.8175,
+      "step": 10404
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.386516795844101e-07,
+      "loss": 0.883,
+      "step": 10405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.376166544871621e-07,
+      "loss": 0.8267,
+      "step": 10406
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.365823272862183e-07,
+      "loss": 0.8679,
+      "step": 10407
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.355486980595172e-07,
+      "loss": 0.8011,
+      "step": 10408
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.345157668849368e-07,
+      "loss": 0.8624,
+      "step": 10409
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.334835338403056e-07,
+      "loss": 0.8287,
+      "step": 10410
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.324519990034018e-07,
+      "loss": 0.8266,
+      "step": 10411
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.314211624519473e-07,
+      "loss": 0.8349,
+      "step": 10412
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.303910242636147e-07,
+      "loss": 0.9067,
+      "step": 10413
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.293615845160196e-07,
+      "loss": 0.7807,
+      "step": 10414
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.283328432867299e-07,
+      "loss": 0.4814,
+      "step": 10415
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.273048006532569e-07,
+      "loss": 0.8733,
+      "step": 10416
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.26277456693063e-07,
+      "loss": 0.8933,
+      "step": 10417
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.25250811483551e-07,
+      "loss": 0.8371,
+      "step": 10418
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.242248651020845e-07,
+      "loss": 0.8337,
+      "step": 10419
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.231996176259581e-07,
+      "loss": 0.8648,
+      "step": 10420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.221750691324247e-07,
+      "loss": 0.8699,
+      "step": 10421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.211512196986803e-07,
+      "loss": 0.8679,
+      "step": 10422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.201280694018708e-07,
+      "loss": 0.8684,
+      "step": 10423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.19105618319087e-07,
+      "loss": 0.8432,
+      "step": 10424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.18083866527367e-07,
+      "loss": 0.8208,
+      "step": 10425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.170628141036984e-07,
+      "loss": 0.8603,
+      "step": 10426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.160424611250139e-07,
+      "loss": 0.7832,
+      "step": 10427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.150228076681954e-07,
+      "loss": 0.5158,
+      "step": 10428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.140038538100714e-07,
+      "loss": 0.8405,
+      "step": 10429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.129855996274148e-07,
+      "loss": 0.8091,
+      "step": 10430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.119680451969524e-07,
+      "loss": 0.9062,
+      "step": 10431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.109511905953504e-07,
+      "loss": 0.8063,
+      "step": 10432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.099350358992285e-07,
+      "loss": 0.8101,
+      "step": 10433
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.089195811851502e-07,
+      "loss": 0.8496,
+      "step": 10434
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.079048265296285e-07,
+      "loss": 0.8639,
+      "step": 10435
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.068907720091211e-07,
+      "loss": 0.8218,
+      "step": 10436
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.05877417700035e-07,
+      "loss": 0.8134,
+      "step": 10437
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.048647636787242e-07,
+      "loss": 0.8624,
+      "step": 10438
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.038528100214892e-07,
+      "loss": 0.8356,
+      "step": 10439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.028415568045799e-07,
+      "loss": 0.8618,
+      "step": 10440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.018310041041854e-07,
+      "loss": 0.8697,
+      "step": 10441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.008211519964558e-07,
+      "loss": 0.8464,
+      "step": 10442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.998120005574749e-07,
+      "loss": 0.8286,
+      "step": 10443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.98803549863285e-07,
+      "loss": 0.6332,
+      "step": 10444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.977957999898643e-07,
+      "loss": 0.8283,
+      "step": 10445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.96788751013151e-07,
+      "loss": 0.8421,
+      "step": 10446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.957824030090165e-07,
+      "loss": 0.8291,
+      "step": 10447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.947767560532925e-07,
+      "loss": 0.9014,
+      "step": 10448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.937718102217461e-07,
+      "loss": 0.837,
+      "step": 10449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.927675655901034e-07,
+      "loss": 0.502,
+      "step": 10450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.917640222340272e-07,
+      "loss": 0.859,
+      "step": 10451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.907611802291325e-07,
+      "loss": 0.8628,
+      "step": 10452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.897590396509823e-07,
+      "loss": 0.8968,
+      "step": 10453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.887576005750851e-07,
+      "loss": 0.8133,
+      "step": 10454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.87756863076896e-07,
+      "loss": 0.8809,
+      "step": 10455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.867568272318171e-07,
+      "loss": 0.8446,
+      "step": 10456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.857574931152e-07,
+      "loss": 0.8097,
+      "step": 10457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.847588608023414e-07,
+      "loss": 0.8592,
+      "step": 10458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.837609303684856e-07,
+      "loss": 0.8657,
+      "step": 10459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.827637018888245e-07,
+      "loss": 0.8434,
+      "step": 10460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.817671754384958e-07,
+      "loss": 0.831,
+      "step": 10461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.807713510925862e-07,
+      "loss": 0.8417,
+      "step": 10462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.797762289261268e-07,
+      "loss": 0.8758,
+      "step": 10463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.787818090140985e-07,
+      "loss": 0.8817,
+      "step": 10464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.777880914314294e-07,
+      "loss": 0.8462,
+      "step": 10465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.767950762529907e-07,
+      "loss": 0.8725,
+      "step": 10466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.758027635536057e-07,
+      "loss": 0.8559,
+      "step": 10467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.748111534080426e-07,
+      "loss": 0.8767,
+      "step": 10468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.73820245891016e-07,
+      "loss": 0.821,
+      "step": 10469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.728300410771871e-07,
+      "loss": 0.8179,
+      "step": 10470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.718405390411676e-07,
+      "loss": 0.8676,
+      "step": 10471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.708517398575121e-07,
+      "loss": 0.8317,
+      "step": 10472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.698636436007256e-07,
+      "loss": 0.8206,
+      "step": 10473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.688762503452551e-07,
+      "loss": 0.5254,
+      "step": 10474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.678895601655012e-07,
+      "loss": 0.8212,
+      "step": 10475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.669035731358075e-07,
+      "loss": 0.8148,
+      "step": 10476
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.659182893304672e-07,
+      "loss": 0.8604,
+      "step": 10477
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.649337088237162e-07,
+      "loss": 0.8585,
+      "step": 10478
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.639498316897419e-07,
+      "loss": 0.8314,
+      "step": 10479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.62966658002675e-07,
+      "loss": 0.8248,
+      "step": 10480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.619841878365974e-07,
+      "loss": 0.8902,
+      "step": 10481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.610024212655364e-07,
+      "loss": 0.8846,
+      "step": 10482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.600213583634596e-07,
+      "loss": 0.8458,
+      "step": 10483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.590409992042957e-07,
+      "loss": 0.8824,
+      "step": 10484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.580613438619044e-07,
+      "loss": 0.847,
+      "step": 10485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.570823924101067e-07,
+      "loss": 0.8538,
+      "step": 10486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.561041449226579e-07,
+      "loss": 0.8617,
+      "step": 10487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.551266014732738e-07,
+      "loss": 0.8321,
+      "step": 10488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.541497621356008e-07,
+      "loss": 0.8445,
+      "step": 10489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.5317362698325e-07,
+      "loss": 0.8479,
+      "step": 10490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.521981960897639e-07,
+      "loss": 0.8698,
+      "step": 10491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.512234695286446e-07,
+      "loss": 0.874,
+      "step": 10492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.50249447373329e-07,
+      "loss": 0.8576,
+      "step": 10493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.492761296972117e-07,
+      "loss": 0.8731,
+      "step": 10494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.483035165736274e-07,
+      "loss": 0.8402,
+      "step": 10495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.473316080758596e-07,
+      "loss": 0.8709,
+      "step": 10496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.463604042771409e-07,
+      "loss": 0.8472,
+      "step": 10497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.453899052506485e-07,
+      "loss": 0.8707,
+      "step": 10498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.444201110695059e-07,
+      "loss": 0.8978,
+      "step": 10499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.434510218067846e-07,
+      "loss": 0.551,
+      "step": 10500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.424826375355042e-07,
+      "loss": 0.8477,
+      "step": 10501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.415149583286296e-07,
+      "loss": 0.8222,
+      "step": 10502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.405479842590723e-07,
+      "loss": 0.8538,
+      "step": 10503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.395817153996909e-07,
+      "loss": 0.8312,
+      "step": 10504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.386161518232925e-07,
+      "loss": 0.8488,
+      "step": 10505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.37651293602628e-07,
+      "loss": 0.8453,
+      "step": 10506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.366871408103992e-07,
+      "loss": 0.8404,
+      "step": 10507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.357236935192512e-07,
+      "loss": 0.8307,
+      "step": 10508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.347609518017761e-07,
+      "loss": 0.8548,
+      "step": 10509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.337989157305168e-07,
+      "loss": 0.8307,
+      "step": 10510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.328375853779578e-07,
+      "loss": 0.852,
+      "step": 10511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.318769608165332e-07,
+      "loss": 0.8357,
+      "step": 10512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.30917042118625e-07,
+      "loss": 0.8762,
+      "step": 10513
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.29957829356559e-07,
+      "loss": 0.8332,
+      "step": 10514
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.289993226026114e-07,
+      "loss": 0.8627,
+      "step": 10515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.280415219289981e-07,
+      "loss": 0.8655,
+      "step": 10516
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.270844274078947e-07,
+      "loss": 0.8294,
+      "step": 10517
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.261280391114077e-07,
+      "loss": 0.8116,
+      "step": 10518
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.251723571116031e-07,
+      "loss": 0.8107,
+      "step": 10519
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.242173814804898e-07,
+      "loss": 0.8549,
+      "step": 10520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.232631122900201e-07,
+      "loss": 0.8581,
+      "step": 10521
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.223095496120968e-07,
+      "loss": 0.8205,
+      "step": 10522
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.213566935185678e-07,
+      "loss": 0.8958,
+      "step": 10523
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2040454408123e-07,
+      "loss": 0.8898,
+      "step": 10524
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.194531013718219e-07,
+      "loss": 0.7969,
+      "step": 10525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.185023654620381e-07,
+      "loss": 0.8669,
+      "step": 10526
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.175523364235059e-07,
+      "loss": 0.8146,
+      "step": 10527
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.166030143278157e-07,
+      "loss": 0.8933,
+      "step": 10528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.156543992464892e-07,
+      "loss": 0.8561,
+      "step": 10529
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.147064912510093e-07,
+      "loss": 0.8253,
+      "step": 10530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.137592904127909e-07,
+      "loss": 0.8369,
+      "step": 10531
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.128127968032105e-07,
+      "loss": 0.8485,
+      "step": 10532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.118670104935765e-07,
+      "loss": 0.8199,
+      "step": 10533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.109219315551584e-07,
+      "loss": 0.8587,
+      "step": 10534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.099775600591606e-07,
+      "loss": 0.8232,
+      "step": 10535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.090338960767417e-07,
+      "loss": 0.4741,
+      "step": 10536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.080909396790025e-07,
+      "loss": 0.8357,
+      "step": 10537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.071486909369928e-07,
+      "loss": 0.8265,
+      "step": 10538
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.062071499217081e-07,
+      "loss": 0.841,
+      "step": 10539
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.052663167040929e-07,
+      "loss": 0.878,
+      "step": 10540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.043261913550347e-07,
+      "loss": 0.8254,
+      "step": 10541
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.033867739453703e-07,
+      "loss": 0.8193,
+      "step": 10542
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.02448064545883e-07,
+      "loss": 0.5426,
+      "step": 10543
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.015100632272997e-07,
+      "loss": 0.8096,
+      "step": 10544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.005727700602992e-07,
+      "loss": 0.8567,
+      "step": 10545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.99636185115502e-07,
+      "loss": 0.8388,
+      "step": 10546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.98700308463479e-07,
+      "loss": 0.8383,
+      "step": 10547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.977651401747442e-07,
+      "loss": 0.8869,
+      "step": 10548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.96830680319761e-07,
+      "loss": 0.8774,
+      "step": 10549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.958969289689387e-07,
+      "loss": 0.8229,
+      "step": 10550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.949638861926333e-07,
+      "loss": 0.8491,
+      "step": 10551
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.940315520611462e-07,
+      "loss": 0.7599,
+      "step": 10552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.93099926644728e-07,
+      "loss": 0.8176,
+      "step": 10553
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.921690100135713e-07,
+      "loss": 0.8476,
+      "step": 10554
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.912388022378212e-07,
+      "loss": 0.8603,
+      "step": 10555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.903093033875651e-07,
+      "loss": 0.8683,
+      "step": 10556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.89380513532839e-07,
+      "loss": 0.8796,
+      "step": 10557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.884524327436214e-07,
+      "loss": 0.8454,
+      "step": 10558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.875250610898475e-07,
+      "loss": 0.8666,
+      "step": 10559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.865983986413848e-07,
+      "loss": 0.8214,
+      "step": 10560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.856724454680607e-07,
+      "loss": 0.8471,
+      "step": 10561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.847472016396394e-07,
+      "loss": 0.8216,
+      "step": 10562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.838226672258374e-07,
+      "loss": 0.8326,
+      "step": 10563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.828988422963167e-07,
+      "loss": 0.8758,
+      "step": 10564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.819757269206839e-07,
+      "loss": 0.8578,
+      "step": 10565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.810533211684954e-07,
+      "loss": 0.826,
+      "step": 10566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.80131625109247e-07,
+      "loss": 0.8597,
+      "step": 10567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.792106388123931e-07,
+      "loss": 0.8021,
+      "step": 10568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.782903623473202e-07,
+      "loss": 0.8574,
+      "step": 10569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.773707957833752e-07,
+      "loss": 0.8758,
+      "step": 10570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.764519391898393e-07,
+      "loss": 0.8883,
+      "step": 10571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.755337926359528e-07,
+      "loss": 0.8528,
+      "step": 10572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.74616356190888e-07,
+      "loss": 0.8253,
+      "step": 10573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.736996299237773e-07,
+      "loss": 0.8798,
+      "step": 10574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.727836139036902e-07,
+      "loss": 0.8583,
+      "step": 10575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.718683081996491e-07,
+      "loss": 0.7969,
+      "step": 10576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.709537128806164e-07,
+      "loss": 0.8589,
+      "step": 10577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.700398280155062e-07,
+      "loss": 0.8023,
+      "step": 10578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.691266536731776e-07,
+      "loss": 0.895,
+      "step": 10579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.682141899224347e-07,
+      "loss": 0.7994,
+      "step": 10580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.673024368320313e-07,
+      "loss": 0.8463,
+      "step": 10581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.663913944706634e-07,
+      "loss": 0.8644,
+      "step": 10582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.654810629069762e-07,
+      "loss": 0.7734,
+      "step": 10583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.645714422095627e-07,
+      "loss": 0.8731,
+      "step": 10584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.63662532446958e-07,
+      "loss": 0.8555,
+      "step": 10585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.627543336876473e-07,
+      "loss": 0.8398,
+      "step": 10586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.618468460000603e-07,
+      "loss": 0.8763,
+      "step": 10587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.609400694525757e-07,
+      "loss": 0.8673,
+      "step": 10588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.600340041135133e-07,
+      "loss": 0.8522,
+      "step": 10589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.591286500511461e-07,
+      "loss": 0.8519,
+      "step": 10590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.582240073336887e-07,
+      "loss": 0.8476,
+      "step": 10591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.573200760293029e-07,
+      "loss": 0.8345,
+      "step": 10592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.564168562060989e-07,
+      "loss": 0.8319,
+      "step": 10593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.555143479321312e-07,
+      "loss": 0.823,
+      "step": 10594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.546125512754019e-07,
+      "loss": 0.8666,
+      "step": 10595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.537114663038579e-07,
+      "loss": 0.8121,
+      "step": 10596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.528110930853947e-07,
+      "loss": 0.9046,
+      "step": 10597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.519114316878538e-07,
+      "loss": 0.8345,
+      "step": 10598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.510124821790208e-07,
+      "loss": 0.8623,
+      "step": 10599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.501142446266283e-07,
+      "loss": 0.8911,
+      "step": 10600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4921671909836e-07,
+      "loss": 0.8381,
+      "step": 10601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.48319905661836e-07,
+      "loss": 0.8674,
+      "step": 10602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.474238043846369e-07,
+      "loss": 0.8693,
+      "step": 10603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.465284153342731e-07,
+      "loss": 0.8397,
+      "step": 10604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.456337385782173e-07,
+      "loss": 0.8724,
+      "step": 10605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.447397741838734e-07,
+      "loss": 0.8413,
+      "step": 10606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.438465222186063e-07,
+      "loss": 0.8374,
+      "step": 10607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.429539827497188e-07,
+      "loss": 0.8683,
+      "step": 10608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.420621558444561e-07,
+      "loss": 0.8426,
+      "step": 10609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.411710415700222e-07,
+      "loss": 0.8188,
+      "step": 10610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.402806399935545e-07,
+      "loss": 0.8668,
+      "step": 10611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.393909511821471e-07,
+      "loss": 0.8439,
+      "step": 10612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.385019752028309e-07,
+      "loss": 0.8174,
+      "step": 10613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.376137121225933e-07,
+      "loss": 0.8979,
+      "step": 10614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.367261620083575e-07,
+      "loss": 0.875,
+      "step": 10615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.358393249270022e-07,
+      "loss": 0.8923,
+      "step": 10616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.34953200945344e-07,
+      "loss": 0.8248,
+      "step": 10617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.340677901301561e-07,
+      "loss": 0.8654,
+      "step": 10618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.331830925481452e-07,
+      "loss": 0.8633,
+      "step": 10619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.322991082659745e-07,
+      "loss": 0.8472,
+      "step": 10620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.314158373502498e-07,
+      "loss": 0.8114,
+      "step": 10621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.305332798675222e-07,
+      "loss": 0.8625,
+      "step": 10622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.296514358842919e-07,
+      "loss": 0.8545,
+      "step": 10623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.287703054670012e-07,
+      "loss": 0.8754,
+      "step": 10624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.278898886820427e-07,
+      "loss": 0.8274,
+      "step": 10625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.270101855957521e-07,
+      "loss": 0.8414,
+      "step": 10626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.261311962744142e-07,
+      "loss": 0.8823,
+      "step": 10627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.252529207842572e-07,
+      "loss": 0.8478,
+      "step": 10628
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.24375359191458e-07,
+      "loss": 0.8362,
+      "step": 10629
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.234985115621372e-07,
+      "loss": 0.843,
+      "step": 10630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.226223779623651e-07,
+      "loss": 0.86,
+      "step": 10631
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.217469584581536e-07,
+      "loss": 0.8692,
+      "step": 10632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.208722531154653e-07,
+      "loss": 0.8762,
+      "step": 10633
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.199982620002064e-07,
+      "loss": 0.8519,
+      "step": 10634
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.191249851782287e-07,
+      "loss": 0.7945,
+      "step": 10635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.18252422715333e-07,
+      "loss": 0.8582,
+      "step": 10636
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.173805746772631e-07,
+      "loss": 0.887,
+      "step": 10637
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.165094411297111e-07,
+      "loss": 0.8257,
+      "step": 10638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.156390221383145e-07,
+      "loss": 0.8138,
+      "step": 10639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.147693177686586e-07,
+      "loss": 0.8782,
+      "step": 10640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.13900328086272e-07,
+      "loss": 0.7918,
+      "step": 10641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.13032053156628e-07,
+      "loss": 0.8266,
+      "step": 10642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.121644930451552e-07,
+      "loss": 0.5887,
+      "step": 10643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.11297647817216e-07,
+      "loss": 0.8727,
+      "step": 10644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.104315175381291e-07,
+      "loss": 0.8761,
+      "step": 10645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.095661022731513e-07,
+      "loss": 0.826,
+      "step": 10646
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.087014020874936e-07,
+      "loss": 0.8712,
+      "step": 10647
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.07837417046304e-07,
+      "loss": 0.8411,
+      "step": 10648
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.069741472146872e-07,
+      "loss": 0.8374,
+      "step": 10649
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.061115926576843e-07,
+      "loss": 0.8542,
+      "step": 10650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.052497534402867e-07,
+      "loss": 0.8587,
+      "step": 10651
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.043886296274347e-07,
+      "loss": 0.8435,
+      "step": 10652
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.035282212840065e-07,
+      "loss": 0.86,
+      "step": 10653
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.026685284748378e-07,
+      "loss": 0.9038,
+      "step": 10654
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.018095512646981e-07,
+      "loss": 0.808,
+      "step": 10655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.009512897183156e-07,
+      "loss": 0.8262,
+      "step": 10656
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.000937439003517e-07,
+      "loss": 0.8041,
+      "step": 10657
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.99236913875426e-07,
+      "loss": 0.8518,
+      "step": 10658
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.983807997080925e-07,
+      "loss": 0.8827,
+      "step": 10659
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.975254014628628e-07,
+      "loss": 0.8313,
+      "step": 10660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.966707192041853e-07,
+      "loss": 0.8624,
+      "step": 10661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.958167529964586e-07,
+      "loss": 0.8751,
+      "step": 10662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.949635029040279e-07,
+      "loss": 0.8269,
+      "step": 10663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.941109689911816e-07,
+      "loss": 0.8515,
+      "step": 10664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.932591513221586e-07,
+      "loss": 0.8554,
+      "step": 10665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.924080499611384e-07,
+      "loss": 0.8915,
+      "step": 10666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.915576649722509e-07,
+      "loss": 0.8272,
+      "step": 10667
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.907079964195693e-07,
+      "loss": 0.5073,
+      "step": 10668
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.898590443671147e-07,
+      "loss": 0.862,
+      "step": 10669
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.890108088788537e-07,
+      "loss": 0.8763,
+      "step": 10670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.881632900186983e-07,
+      "loss": 0.8504,
+      "step": 10671
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.873164878505066e-07,
+      "loss": 0.929,
+      "step": 10672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.864704024380818e-07,
+      "loss": 0.8163,
+      "step": 10673
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.856250338451763e-07,
+      "loss": 0.8738,
+      "step": 10674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.847803821354858e-07,
+      "loss": 0.8611,
+      "step": 10675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.839364473726515e-07,
+      "loss": 0.8289,
+      "step": 10676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.83093229620264e-07,
+      "loss": 0.822,
+      "step": 10677
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.822507289418543e-07,
+      "loss": 0.5766,
+      "step": 10678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.814089454009053e-07,
+      "loss": 0.8993,
+      "step": 10679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.805678790608415e-07,
+      "loss": 0.8772,
+      "step": 10680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.797275299850368e-07,
+      "loss": 0.8718,
+      "step": 10681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.788878982368072e-07,
+      "loss": 0.876,
+      "step": 10682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7804898387942e-07,
+      "loss": 0.8587,
+      "step": 10683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7721078697608e-07,
+      "loss": 0.8858,
+      "step": 10684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.763733075899479e-07,
+      "loss": 0.8457,
+      "step": 10685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.755365457841221e-07,
+      "loss": 0.8222,
+      "step": 10686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.747005016216555e-07,
+      "loss": 0.7986,
+      "step": 10687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7386517516553434e-07,
+      "loss": 0.8468,
+      "step": 10688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7303056647870605e-07,
+      "loss": 0.8834,
+      "step": 10689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7219667562404926e-07,
+      "loss": 0.8195,
+      "step": 10690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.713635026644014e-07,
+      "loss": 0.8622,
+      "step": 10691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705310476625369e-07,
+      "loss": 0.8391,
+      "step": 10692
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6969931068117867e-07,
+      "loss": 0.5023,
+      "step": 10693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6886829178299676e-07,
+      "loss": 0.774,
+      "step": 10694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6803799103060544e-07,
+      "loss": 0.8097,
+      "step": 10695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.67208408486568e-07,
+      "loss": 0.8509,
+      "step": 10696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6637954421338783e-07,
+      "loss": 0.8489,
+      "step": 10697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.655513982735216e-07,
+      "loss": 0.8867,
+      "step": 10698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.647239707293638e-07,
+      "loss": 0.8924,
+      "step": 10699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6389726164326354e-07,
+      "loss": 0.8364,
+      "step": 10700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.630712710775054e-07,
+      "loss": 0.8943,
+      "step": 10701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.622459990943318e-07,
+      "loss": 0.8286,
+      "step": 10702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6142144575592075e-07,
+      "loss": 0.848,
+      "step": 10703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.605976111244015e-07,
+      "loss": 0.8947,
+      "step": 10704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.597744952618477e-07,
+      "loss": 0.8657,
+      "step": 10705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.589520982302786e-07,
+      "loss": 0.8329,
+      "step": 10706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.581304200916603e-07,
+      "loss": 0.8428,
+      "step": 10707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.573094609079032e-07,
+      "loss": 0.4529,
+      "step": 10708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5648922074086553e-07,
+      "loss": 0.8572,
+      "step": 10709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.556696996523502e-07,
+      "loss": 0.8689,
+      "step": 10710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.556696996523502e-07,
+      "loss": 0.837,
+      "step": 10711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.548508977041044e-07,
+      "loss": 0.8462,
+      "step": 10712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5403281495782437e-07,
+      "loss": 0.8554,
+      "step": 10713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.532154514751497e-07,
+      "loss": 0.8557,
+      "step": 10714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.523988073176666e-07,
+      "loss": 0.823,
+      "step": 10715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.515828825469082e-07,
+      "loss": 0.8835,
+      "step": 10716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5076767722435075e-07,
+      "loss": 0.8258,
+      "step": 10717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.499531914114186e-07,
+      "loss": 0.8275,
+      "step": 10718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4913942516948027e-07,
+      "loss": 0.8316,
+      "step": 10719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.483263785598524e-07,
+      "loss": 0.8455,
+      "step": 10720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.475140516437959e-07,
+      "loss": 0.8313,
+      "step": 10721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.467024444825152e-07,
+      "loss": 0.8461,
+      "step": 10722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.458915571371647e-07,
+      "loss": 0.8447,
+      "step": 10723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.450813896688433e-07,
+      "loss": 0.8563,
+      "step": 10724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4427194213859216e-07,
+      "loss": 0.8684,
+      "step": 10725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4346321460740583e-07,
+      "loss": 0.8813,
+      "step": 10726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.426552071362133e-07,
+      "loss": 0.8326,
+      "step": 10727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.418479197859016e-07,
+      "loss": 0.822,
+      "step": 10728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4104135261729296e-07,
+      "loss": 0.8389,
+      "step": 10729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.402355056911656e-07,
+      "loss": 0.4716,
+      "step": 10730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3943037906823195e-07,
+      "loss": 0.8305,
+      "step": 10731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.386259728091613e-07,
+      "loss": 0.8564,
+      "step": 10732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3782228697455965e-07,
+      "loss": 0.8246,
+      "step": 10733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.370193216249874e-07,
+      "loss": 0.8521,
+      "step": 10734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3621707682094063e-07,
+      "loss": 0.8519,
+      "step": 10735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.354155526228687e-07,
+      "loss": 0.8036,
+      "step": 10736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3461474909116454e-07,
+      "loss": 0.5451,
+      "step": 10737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.338146662861664e-07,
+      "loss": 0.862,
+      "step": 10738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.330153042681573e-07,
+      "loss": 0.8242,
+      "step": 10739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3221666309736787e-07,
+      "loss": 0.8749,
+      "step": 10740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3141874283397665e-07,
+      "loss": 0.8634,
+      "step": 10741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.306215435380989e-07,
+      "loss": 0.8927,
+      "step": 10742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.298250652698077e-07,
+      "loss": 0.867,
+      "step": 10743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2902930808910946e-07,
+      "loss": 0.8249,
+      "step": 10744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2823427205596845e-07,
+      "loss": 0.8607,
+      "step": 10745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.274399572302845e-07,
+      "loss": 0.7797,
+      "step": 10746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.266463636719087e-07,
+      "loss": 0.8852,
+      "step": 10747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2585349144063424e-07,
+      "loss": 0.8671,
+      "step": 10748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2506134059620453e-07,
+      "loss": 0.8618,
+      "step": 10749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.242699111983051e-07,
+      "loss": 0.8302,
+      "step": 10750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.234792033065682e-07,
+      "loss": 0.857,
+      "step": 10751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2268921698057076e-07,
+      "loss": 0.8506,
+      "step": 10752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2189995227983726e-07,
+      "loss": 0.8225,
+      "step": 10753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2111140926383686e-07,
+      "loss": 0.8167,
+      "step": 10754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2032358799198314e-07,
+      "loss": 0.8396,
+      "step": 10755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.195364885236375e-07,
+      "loss": 0.816,
+      "step": 10756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1875011091810467e-07,
+      "loss": 0.8197,
+      "step": 10757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1796445523463626e-07,
+      "loss": 0.4776,
+      "step": 10758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1717952153243034e-07,
+      "loss": 0.9156,
+      "step": 10759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.163953098706286e-07,
+      "loss": 0.8332,
+      "step": 10760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1561182030832035e-07,
+      "loss": 0.7981,
+      "step": 10761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1482905290453846e-07,
+      "loss": 0.8687,
+      "step": 10762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.140470077182634e-07,
+      "loss": 0.8574,
+      "step": 10763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.132656848084193e-07,
+      "loss": 0.8989,
+      "step": 10764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.124850842338779e-07,
+      "loss": 0.8499,
+      "step": 10765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1170520605345545e-07,
+      "loss": 0.8469,
+      "step": 10766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.109260503259105e-07,
+      "loss": 0.8641,
+      "step": 10767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.10147617109955e-07,
+      "loss": 0.8233,
+      "step": 10768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.093699064642376e-07,
+      "loss": 0.8249,
+      "step": 10769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.085929184473614e-07,
+      "loss": 0.8309,
+      "step": 10770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.078166531178651e-07,
+      "loss": 0.8203,
+      "step": 10771
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0704111053424314e-07,
+      "loss": 0.819,
+      "step": 10772
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.062662907549264e-07,
+      "loss": 0.8735,
+      "step": 10773
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0549219383830054e-07,
+      "loss": 0.9061,
+      "step": 10774
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0471881984268655e-07,
+      "loss": 0.8717,
+      "step": 10775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.039461688263602e-07,
+      "loss": 0.823,
+      "step": 10776
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.031742408475359e-07,
+      "loss": 0.9192,
+      "step": 10777
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0240303596437713e-07,
+      "loss": 0.5133,
+      "step": 10778
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.01632554234993e-07,
+      "loss": 0.8509,
+      "step": 10779
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0086279571743715e-07,
+      "loss": 0.4893,
+      "step": 10780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0009376046970747e-07,
+      "loss": 0.8255,
+      "step": 10781
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9932544854975e-07,
+      "loss": 0.856,
+      "step": 10782
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.985578600154549e-07,
+      "loss": 0.8802,
+      "step": 10783
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9779099492465724e-07,
+      "loss": 0.8144,
+      "step": 10784
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.970248533351406e-07,
+      "loss": 0.82,
+      "step": 10785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9625943530462787e-07,
+      "loss": 0.8773,
+      "step": 10786
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.954947408907961e-07,
+      "loss": 0.8308,
+      "step": 10787
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9473077015125814e-07,
+      "loss": 0.8497,
+      "step": 10788
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.939675231435802e-07,
+      "loss": 0.8463,
+      "step": 10789
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.932049999252707e-07,
+      "loss": 0.8452,
+      "step": 10790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.924432005537826e-07,
+      "loss": 0.8941,
+      "step": 10791
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9168212508651547e-07,
+      "loss": 0.8196,
+      "step": 10792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.909217735808157e-07,
+      "loss": 0.8084,
+      "step": 10793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9016214609397306e-07,
+      "loss": 0.8966,
+      "step": 10794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8940324268322285e-07,
+      "loss": 0.8587,
+      "step": 10795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8864506340574817e-07,
+      "loss": 0.8355,
+      "step": 10796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8788760831867446e-07,
+      "loss": 0.8413,
+      "step": 10797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8713087747907385e-07,
+      "loss": 0.8373,
+      "step": 10798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.863748709439641e-07,
+      "loss": 0.8393,
+      "step": 10799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8561958877030957e-07,
+      "loss": 0.5651,
+      "step": 10800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8486503101501705e-07,
+      "loss": 0.7935,
+      "step": 10801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.841111977349421e-07,
+      "loss": 0.8537,
+      "step": 10802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8335808898688376e-07,
+      "loss": 0.8404,
+      "step": 10803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8260570482758554e-07,
+      "loss": 0.8715,
+      "step": 10804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.818540453137376e-07,
+      "loss": 0.8677,
+      "step": 10805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8110311050197803e-07,
+      "loss": 0.8352,
+      "step": 10806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.803529004488848e-07,
+      "loss": 0.8109,
+      "step": 10807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.796034152109873e-07,
+      "loss": 0.8275,
+      "step": 10808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.788546548447536e-07,
+      "loss": 0.8643,
+      "step": 10809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.781066194066052e-07,
+      "loss": 0.8346,
+      "step": 10810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7735930895289927e-07,
+      "loss": 0.851,
+      "step": 10811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.766127235399486e-07,
+      "loss": 0.828,
+      "step": 10812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7586686322400257e-07,
+      "loss": 0.4633,
+      "step": 10813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.751217280612651e-07,
+      "loss": 0.8455,
+      "step": 10814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7437731810787246e-07,
+      "loss": 0.8732,
+      "step": 10815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7363363341992197e-07,
+      "loss": 0.8524,
+      "step": 10816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.728906740534421e-07,
+      "loss": 0.856,
+      "step": 10817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.721484400644182e-07,
+      "loss": 0.8113,
+      "step": 10818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.714069315087709e-07,
+      "loss": 0.8426,
+      "step": 10819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7066614844237346e-07,
+      "loss": 0.9048,
+      "step": 10820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6992609092104116e-07,
+      "loss": 0.8509,
+      "step": 10821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6918675900053605e-07,
+      "loss": 0.9295,
+      "step": 10822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6844815273656356e-07,
+      "loss": 0.5608,
+      "step": 10823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6771027218477694e-07,
+      "loss": 0.871,
+      "step": 10824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.66973117400774e-07,
+      "loss": 0.8898,
+      "step": 10825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6623668844009585e-07,
+      "loss": 0.8833,
+      "step": 10826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.655009853582314e-07,
+      "loss": 0.8428,
+      "step": 10827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.64766008210613e-07,
+      "loss": 0.8376,
+      "step": 10828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6403175705262307e-07,
+      "loss": 0.8472,
+      "step": 10829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.632982319395806e-07,
+      "loss": 0.8481,
+      "step": 10830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6256543292675584e-07,
+      "loss": 0.8436,
+      "step": 10831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6183336006936577e-07,
+      "loss": 0.8759,
+      "step": 10832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6110201342256734e-07,
+      "loss": 0.836,
+      "step": 10833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.603713930414676e-07,
+      "loss": 0.7937,
+      "step": 10834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5964149898111587e-07,
+      "loss": 0.5294,
+      "step": 10835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.58912331296507e-07,
+      "loss": 0.8458,
+      "step": 10836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.581838900425838e-07,
+      "loss": 0.7739,
+      "step": 10837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.574561752742323e-07,
+      "loss": 0.8277,
+      "step": 10838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.567291870462819e-07,
+      "loss": 0.8417,
+      "step": 10839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5600292541351e-07,
+      "loss": 0.8739,
+      "step": 10840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.552773904306406e-07,
+      "loss": 0.8421,
+      "step": 10841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5455258215233655e-07,
+      "loss": 0.8449,
+      "step": 10842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.538285006332154e-07,
+      "loss": 0.8544,
+      "step": 10843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.531051459278312e-07,
+      "loss": 0.8652,
+      "step": 10844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.52382518090687e-07,
+      "loss": 0.8125,
+      "step": 10845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.516606171762338e-07,
+      "loss": 0.7982,
+      "step": 10846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.509394432388613e-07,
+      "loss": 0.8897,
+      "step": 10847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5021899633291056e-07,
+      "loss": 0.8511,
+      "step": 10848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.494992765126637e-07,
+      "loss": 0.8353,
+      "step": 10849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4878028383235286e-07,
+      "loss": 0.8567,
+      "step": 10850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4806201834614697e-07,
+      "loss": 0.8134,
+      "step": 10851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4734448010817043e-07,
+      "loss": 0.8393,
+      "step": 10852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466276691724846e-07,
+      "loss": 0.9071,
+      "step": 10853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4591158559310167e-07,
+      "loss": 0.8774,
+      "step": 10854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.451962294239741e-07,
+      "loss": 0.8407,
+      "step": 10855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.444816007190055e-07,
+      "loss": 0.8378,
+      "step": 10856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4376769953203715e-07,
+      "loss": 0.851,
+      "step": 10857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.430545259168638e-07,
+      "loss": 0.546,
+      "step": 10858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.423420799272181e-07,
+      "loss": 0.8704,
+      "step": 10859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4163036161678267e-07,
+      "loss": 0.8522,
+      "step": 10860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.409193710391834e-07,
+      "loss": 0.8127,
+      "step": 10861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.402091082479908e-07,
+      "loss": 0.8397,
+      "step": 10862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.39499573296721e-07,
+      "loss": 0.8276,
+      "step": 10863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3879076623883677e-07,
+      "loss": 0.8272,
+      "step": 10864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.380826871277443e-07,
+      "loss": 0.8469,
+      "step": 10865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.373753360167964e-07,
+      "loss": 0.8593,
+      "step": 10866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3666871295928826e-07,
+      "loss": 0.8881,
+      "step": 10867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.35962818008464e-07,
+      "loss": 0.8613,
+      "step": 10868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3525765121750986e-07,
+      "loss": 0.899,
+      "step": 10869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.345532126395579e-07,
+      "loss": 0.8374,
+      "step": 10870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3384950232768775e-07,
+      "loss": 0.8591,
+      "step": 10871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3314652033491933e-07,
+      "loss": 0.8305,
+      "step": 10872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3244426671422246e-07,
+      "loss": 0.8137,
+      "step": 10873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.317427415185104e-07,
+      "loss": 0.8081,
+      "step": 10874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3104194480063967e-07,
+      "loss": 0.8562,
+      "step": 10875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3034187661341476e-07,
+      "loss": 0.8536,
+      "step": 10876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.296425370095824e-07,
+      "loss": 0.7915,
+      "step": 10877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.289439260418381e-07,
+      "loss": 0.8398,
+      "step": 10878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2824604376281874e-07,
+      "loss": 0.8764,
+      "step": 10879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2754889022510784e-07,
+      "loss": 0.8792,
+      "step": 10880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.268524654812355e-07,
+      "loss": 0.8553,
+      "step": 10881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2615676958367424e-07,
+      "loss": 0.8091,
+      "step": 10882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.254618025848455e-07,
+      "loss": 0.8082,
+      "step": 10883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2476756453710843e-07,
+      "loss": 0.8306,
+      "step": 10884
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2407405549277683e-07,
+      "loss": 0.8142,
+      "step": 10885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.233812755040999e-07,
+      "loss": 0.863,
+      "step": 10886
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2268922462328155e-07,
+      "loss": 0.8137,
+      "step": 10887
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.219979029024634e-07,
+      "loss": 0.8258,
+      "step": 10888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2130731039373585e-07,
+      "loss": 0.8525,
+      "step": 10889
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.20617447149133e-07,
+      "loss": 0.7906,
+      "step": 10890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1992831322063324e-07,
+      "loss": 0.8646,
+      "step": 10891
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1923990866016385e-07,
+      "loss": 0.8881,
+      "step": 10892
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1855223351958897e-07,
+      "loss": 0.8171,
+      "step": 10893
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.178652878507293e-07,
+      "loss": 0.8053,
+      "step": 10894
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.171790717053391e-07,
+      "loss": 0.8315,
+      "step": 10895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1649358513512693e-07,
+      "loss": 0.8722,
+      "step": 10896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.158088281917393e-07,
+      "loss": 0.8143,
+      "step": 10897
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1512480092677377e-07,
+      "loss": 0.8806,
+      "step": 10898
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1444150339176693e-07,
+      "loss": 0.8464,
+      "step": 10899
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.137589356382076e-07,
+      "loss": 0.8525,
+      "step": 10900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1307709771752124e-07,
+      "loss": 0.8245,
+      "step": 10901
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1239598968108577e-07,
+      "loss": 0.8474,
+      "step": 10902
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.117156115802178e-07,
+      "loss": 0.8351,
+      "step": 10903
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1103596346618415e-07,
+      "loss": 0.8021,
+      "step": 10904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1035704539019384e-07,
+      "loss": 0.895,
+      "step": 10905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0967885740340266e-07,
+      "loss": 0.9192,
+      "step": 10906
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0900139955690857e-07,
+      "loss": 0.8294,
+      "step": 10907
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.083246719017563e-07,
+      "loss": 0.8642,
+      "step": 10908
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.076486744889373e-07,
+      "loss": 0.8993,
+      "step": 10909
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.069734073693853e-07,
+      "loss": 0.8371,
+      "step": 10910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0629887059397954e-07,
+      "loss": 0.4201,
+      "step": 10911
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.05625064213545e-07,
+      "loss": 0.8553,
+      "step": 10912
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.04951988278851e-07,
+      "loss": 0.8532,
+      "step": 10913
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.042796428406114e-07,
+      "loss": 0.8341,
+      "step": 10914
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0360802794948687e-07,
+      "loss": 0.845,
+      "step": 10915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.029371436560824e-07,
+      "loss": 0.8364,
+      "step": 10916
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.022669900109443e-07,
+      "loss": 0.8586,
+      "step": 10917
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0159756706456987e-07,
+      "loss": 0.8333,
+      "step": 10918
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0092887486739776e-07,
+      "loss": 0.8128,
+      "step": 10919
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.00260913469812e-07,
+      "loss": 0.8572,
+      "step": 10920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.995936829221413e-07,
+      "loss": 0.8329,
+      "step": 10921
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9892718327465873e-07,
+      "loss": 0.8512,
+      "step": 10922
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.982614145775864e-07,
+      "loss": 0.5058,
+      "step": 10923
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.975963768810852e-07,
+      "loss": 0.7997,
+      "step": 10924
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9693207023526625e-07,
+      "loss": 0.8564,
+      "step": 10925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.962684946901806e-07,
+      "loss": 0.8387,
+      "step": 10926
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.956056502958304e-07,
+      "loss": 0.8514,
+      "step": 10927
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9494353710215475e-07,
+      "loss": 0.8607,
+      "step": 10928
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.94282155159048e-07,
+      "loss": 0.8449,
+      "step": 10929
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.936215045163371e-07,
+      "loss": 0.8134,
+      "step": 10930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9296158522380546e-07,
+      "loss": 0.8527,
+      "step": 10931
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.923023973311745e-07,
+      "loss": 0.7946,
+      "step": 10932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.916439408881111e-07,
+      "loss": 0.8744,
+      "step": 10933
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9098621594423116e-07,
+      "loss": 0.8189,
+      "step": 10934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9032922254908833e-07,
+      "loss": 0.8711,
+      "step": 10935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.896729607521898e-07,
+      "loss": 0.8287,
+      "step": 10936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8901743060297917e-07,
+      "loss": 0.8046,
+      "step": 10937
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.883626321508526e-07,
+      "loss": 0.883,
+      "step": 10938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8770856544514393e-07,
+      "loss": 0.9019,
+      "step": 10939
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.870552305351382e-07,
+      "loss": 0.849,
+      "step": 10940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8640262747006043e-07,
+      "loss": 0.8388,
+      "step": 10941
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8575075629908465e-07,
+      "loss": 0.5737,
+      "step": 10942
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8509961707132496e-07,
+      "loss": 0.8979,
+      "step": 10943
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8444920983584647e-07,
+      "loss": 0.8741,
+      "step": 10944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8379953464165334e-07,
+      "loss": 0.8227,
+      "step": 10945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.831505915376964e-07,
+      "loss": 0.8166,
+      "step": 10946
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8250238057287326e-07,
+      "loss": 0.8745,
+      "step": 10947
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.818549017960237e-07,
+      "loss": 0.8551,
+      "step": 10948
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.812081552559343e-07,
+      "loss": 0.8451,
+      "step": 10949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8056214100133593e-07,
+      "loss": 0.8607,
+      "step": 10950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7991685908090316e-07,
+      "loss": 0.8357,
+      "step": 10951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.79272309543257e-07,
+      "loss": 0.8255,
+      "step": 10952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.786284924369631e-07,
+      "loss": 0.8367,
+      "step": 10953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.779854078105304e-07,
+      "loss": 0.8635,
+      "step": 10954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.773430557124135e-07,
+      "loss": 0.8822,
+      "step": 10955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7670143619101255e-07,
+      "loss": 0.8408,
+      "step": 10956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.760605492946722e-07,
+      "loss": 0.8364,
+      "step": 10957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.754203950716805e-07,
+      "loss": 0.8552,
+      "step": 10958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7478097357027336e-07,
+      "loss": 0.8323,
+      "step": 10959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.741422848386266e-07,
+      "loss": 0.8501,
+      "step": 10960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.735043289248662e-07,
+      "loss": 0.849,
+      "step": 10961
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7286710587705933e-07,
+      "loss": 0.8392,
+      "step": 10962
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7223061574321975e-07,
+      "loss": 0.8255,
+      "step": 10963
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.715948585713046e-07,
+      "loss": 0.8126,
+      "step": 10964
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7095983440921683e-07,
+      "loss": 0.8803,
+      "step": 10965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7032554330480464e-07,
+      "loss": 0.8535,
+      "step": 10966
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.696919853058588e-07,
+      "loss": 0.8834,
+      "step": 10967
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.690591604601167e-07,
+      "loss": 0.8978,
+      "step": 10968
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6842706881526125e-07,
+      "loss": 0.8747,
+      "step": 10969
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.677957104189155e-07,
+      "loss": 0.8512,
+      "step": 10970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.671650853186558e-07,
+      "loss": 0.849,
+      "step": 10971
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.665351935619931e-07,
+      "loss": 0.8712,
+      "step": 10972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6590603519639157e-07,
+      "loss": 0.851,
+      "step": 10973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.652776102692545e-07,
+      "loss": 0.808,
+      "step": 10974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.646499188279328e-07,
+      "loss": 0.8173,
+      "step": 10975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6402296091972314e-07,
+      "loss": 0.8415,
+      "step": 10976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6339673659186216e-07,
+      "loss": 0.8273,
+      "step": 10977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6277124589153657e-07,
+      "loss": 0.8636,
+      "step": 10978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6214648886587315e-07,
+      "loss": 0.8565,
+      "step": 10979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6152246556194863e-07,
+      "loss": 0.8712,
+      "step": 10980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.608991760267776e-07,
+      "loss": 0.8375,
+      "step": 10981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6027662030732814e-07,
+      "loss": 0.8415,
+      "step": 10982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.596547984505027e-07,
+      "loss": 0.8291,
+      "step": 10983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.590337105031604e-07,
+      "loss": 0.8303,
+      "step": 10984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5841335651209166e-07,
+      "loss": 0.8742,
+      "step": 10985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.577937365240446e-07,
+      "loss": 0.8681,
+      "step": 10986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5717485058570304e-07,
+      "loss": 0.8648,
+      "step": 10987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5655669874369736e-07,
+      "loss": 0.8399,
+      "step": 10988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5593928104460596e-07,
+      "loss": 0.8537,
+      "step": 10989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5532259753494825e-07,
+      "loss": 0.8778,
+      "step": 10990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.547066482611904e-07,
+      "loss": 0.8261,
+      "step": 10991
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5409143326974196e-07,
+      "loss": 0.8488,
+      "step": 10992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5347695260695805e-07,
+      "loss": 0.8146,
+      "step": 10993
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.528632063191394e-07,
+      "loss": 0.84,
+      "step": 10994
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5225019445252795e-07,
+      "loss": 0.828,
+      "step": 10995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5163791705331343e-07,
+      "loss": 0.8293,
+      "step": 10996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5102637416762996e-07,
+      "loss": 0.4812,
+      "step": 10997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5041556584155523e-07,
+      "loss": 0.8786,
+      "step": 10998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4980549212111236e-07,
+      "loss": 0.8533,
+      "step": 10999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.491961530522691e-07,
+      "loss": 0.9,
+      "step": 11000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.485875486809364e-07,
+      "loss": 0.8719,
+      "step": 11001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4797967905297093e-07,
+      "loss": 0.8485,
+      "step": 11002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.47372544214175e-07,
+      "loss": 0.8625,
+      "step": 11003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.467661442102953e-07,
+      "loss": 0.8191,
+      "step": 11004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.461604790870209e-07,
+      "loss": 0.7934,
+      "step": 11005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4555554888998745e-07,
+      "loss": 0.9078,
+      "step": 11006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.449513536647752e-07,
+      "loss": 0.8051,
+      "step": 11007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.443478934569088e-07,
+      "loss": 0.8999,
+      "step": 11008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4374516831185857e-07,
+      "loss": 0.8856,
+      "step": 11009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4314317827503375e-07,
+      "loss": 0.8803,
+      "step": 11010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4254192339179915e-07,
+      "loss": 0.8742,
+      "step": 11011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.419414037074519e-07,
+      "loss": 0.7736,
+      "step": 11012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.413416192672435e-07,
+      "loss": 0.8515,
+      "step": 11013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.407425701163635e-07,
+      "loss": 0.8797,
+      "step": 11014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.401442562999523e-07,
+      "loss": 0.5182,
+      "step": 11015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3954667786308614e-07,
+      "loss": 0.8027,
+      "step": 11016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3894983485079683e-07,
+      "loss": 0.8764,
+      "step": 11017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.383537273080505e-07,
+      "loss": 0.8715,
+      "step": 11018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3775835527976243e-07,
+      "loss": 0.8694,
+      "step": 11019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3716371881079558e-07,
+      "loss": 0.8967,
+      "step": 11020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.365698179459497e-07,
+      "loss": 0.8665,
+      "step": 11021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3597665272997893e-07,
+      "loss": 0.8504,
+      "step": 11022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3538422320757204e-07,
+      "loss": 0.8571,
+      "step": 11023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3479252942337106e-07,
+      "loss": 0.868,
+      "step": 11024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3420157142195365e-07,
+      "loss": 0.8234,
+      "step": 11025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3361134924785313e-07,
+      "loss": 0.8408,
+      "step": 11026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3302186294553497e-07,
+      "loss": 0.8522,
+      "step": 11027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3243311255942035e-07,
+      "loss": 0.8595,
+      "step": 11028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3184509813386713e-07,
+      "loss": 0.8266,
+      "step": 11029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3125781971318207e-07,
+      "loss": 0.8566,
+      "step": 11030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.306712773416142e-07,
+      "loss": 0.8239,
+      "step": 11031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3008547106335822e-07,
+      "loss": 0.891,
+      "step": 11032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2950040092255322e-07,
+      "loss": 0.8766,
+      "step": 11033
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2891606696328283e-07,
+      "loss": 0.8326,
+      "step": 11034
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2833246922957408e-07,
+      "loss": 0.8133,
+      "step": 11035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2774960776540178e-07,
+      "loss": 0.8472,
+      "step": 11036
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.271674826146808e-07,
+      "loss": 0.8242,
+      "step": 11037
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2658609382127384e-07,
+      "loss": 0.8527,
+      "step": 11038
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2600544142898585e-07,
+      "loss": 0.8461,
+      "step": 11039
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.254255254815696e-07,
+      "loss": 0.8584,
+      "step": 11040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2484634602271794e-07,
+      "loss": 0.8306,
+      "step": 11041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.242679030960726e-07,
+      "loss": 0.8623,
+      "step": 11042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2369019674521541e-07,
+      "loss": 0.8204,
+      "step": 11043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.231132270136771e-07,
+      "loss": 0.8507,
+      "step": 11044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2253699394493066e-07,
+      "loss": 0.8242,
+      "step": 11045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.219614975823914e-07,
+      "loss": 0.8742,
+      "step": 11046
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2138673796942457e-07,
+      "loss": 0.8155,
+      "step": 11047
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2081271514933557e-07,
+      "loss": 0.7923,
+      "step": 11048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2023942916537422e-07,
+      "loss": 0.8827,
+      "step": 11049
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.196668800607382e-07,
+      "loss": 0.8369,
+      "step": 11050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.190950678785675e-07,
+      "loss": 0.8507,
+      "step": 11051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 0.8371,
+      "step": 11052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.179536544538996e-07,
+      "loss": 0.9323,
+      "step": 11053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1738405329740586e-07,
+      "loss": 0.8866,
+      "step": 11054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1681518923538315e-07,
+      "loss": 0.8794,
+      "step": 11055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1624706231068936e-07,
+      "loss": 0.8395,
+      "step": 11056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.156796725661381e-07,
+      "loss": 0.852,
+      "step": 11057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.15113020044474e-07,
+      "loss": 0.8873,
+      "step": 11058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1454710478839736e-07,
+      "loss": 0.8571,
+      "step": 11059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.139819268405463e-07,
+      "loss": 0.8744,
+      "step": 11060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1341748624350678e-07,
+      "loss": 0.8357,
+      "step": 11061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1285378303980585e-07,
+      "loss": 0.8495,
+      "step": 11062
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1229081727191846e-07,
+      "loss": 0.7865,
+      "step": 11063
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.11728588982264e-07,
+      "loss": 0.8611,
+      "step": 11064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.111670982132008e-07,
+      "loss": 0.8537,
+      "step": 11065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1060634500704168e-07,
+      "loss": 0.8141,
+      "step": 11066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1004632940603176e-07,
+      "loss": 0.8645,
+      "step": 11067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0948705145237168e-07,
+      "loss": 0.8193,
+      "step": 11068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0892851118819778e-07,
+      "loss": 0.8345,
+      "step": 11069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.083707086555975e-07,
+      "loss": 0.8284,
+      "step": 11070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.078136438965983e-07,
+      "loss": 0.7958,
+      "step": 11071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.072573169531744e-07,
+      "loss": 0.8241,
+      "step": 11072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.067017278672423e-07,
+      "loss": 0.8753,
+      "step": 11073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0614687668066403e-07,
+      "loss": 0.8338,
+      "step": 11074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0559276343524838e-07,
+      "loss": 0.8524,
+      "step": 11075
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0503938817274416e-07,
+      "loss": 0.8916,
+      "step": 11076
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0448675093484805e-07,
+      "loss": 0.8334,
+      "step": 11077
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0393485176320005e-07,
+      "loss": 0.8695,
+      "step": 11078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0338369069938246e-07,
+      "loss": 0.8427,
+      "step": 11079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.028332677849254e-07,
+      "loss": 0.8798,
+      "step": 11080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0228358306130124e-07,
+      "loss": 0.8801,
+      "step": 11081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0173463656992797e-07,
+      "loss": 0.8523,
+      "step": 11082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0118642835216584e-07,
+      "loss": 0.8325,
+      "step": 11083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.006389584493218e-07,
+      "loss": 0.8195,
+      "step": 11084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.000922269026473e-07,
+      "loss": 0.8176,
+      "step": 11085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9954623375333493e-07,
+      "loss": 0.8595,
+      "step": 11086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9900097904252513e-07,
+      "loss": 0.823,
+      "step": 11087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.984564628113017e-07,
+      "loss": 0.8364,
+      "step": 11088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9791268510069184e-07,
+      "loss": 0.8984,
+      "step": 11089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9736964595166829e-07,
+      "loss": 0.8327,
+      "step": 11090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9682734540514724e-07,
+      "loss": 0.636,
+      "step": 11091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9628578350198933e-07,
+      "loss": 0.8238,
+      "step": 11092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9574496028300194e-07,
+      "loss": 0.8359,
+      "step": 11093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9520487578893022e-07,
+      "loss": 0.8103,
+      "step": 11094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9466553006047383e-07,
+      "loss": 0.8242,
+      "step": 11095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9412692313826586e-07,
+      "loss": 0.8705,
+      "step": 11096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9358905506289271e-07,
+      "loss": 0.8396,
+      "step": 11097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9305192587487753e-07,
+      "loss": 0.8326,
+      "step": 11098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.925155356146957e-07,
+      "loss": 0.825,
+      "step": 11099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9197988432276048e-07,
+      "loss": 0.8181,
+      "step": 11100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.914449720394329e-07,
+      "loss": 0.84,
+      "step": 11101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.909107988050163e-07,
+      "loss": 0.8133,
+      "step": 11102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9037736465975842e-07,
+      "loss": 0.8583,
+      "step": 11103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8984466964385384e-07,
+      "loss": 0.84,
+      "step": 11104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8931271379743931e-07,
+      "loss": 0.8967,
+      "step": 11105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8878149716059613e-07,
+      "loss": 0.8674,
+      "step": 11106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8825101977334891e-07,
+      "loss": 0.8659,
+      "step": 11107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8772128167567128e-07,
+      "loss": 0.829,
+      "step": 11108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8719228290747348e-07,
+      "loss": 0.8408,
+      "step": 11109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8666402350861701e-07,
+      "loss": 0.8013,
+      "step": 11110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8613650351890222e-07,
+      "loss": 0.8722,
+      "step": 11111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8560972297807956e-07,
+      "loss": 0.8827,
+      "step": 11112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8508368192583838e-07,
+      "loss": 0.8466,
+      "step": 11113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8455838040181362e-07,
+      "loss": 0.865,
+      "step": 11114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.840338184455881e-07,
+      "loss": 0.8588,
+      "step": 11115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8350999609668462e-07,
+      "loss": 0.8669,
+      "step": 11116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8298691339457164e-07,
+      "loss": 0.8642,
+      "step": 11117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8246457037866316e-07,
+      "loss": 0.8209,
+      "step": 11118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8194296708831548e-07,
+      "loss": 0.8138,
+      "step": 11119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.814221035628294e-07,
+      "loss": 0.8065,
+      "step": 11120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8090197984145242e-07,
+      "loss": 0.8385,
+      "step": 11121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8038259596337316e-07,
+      "loss": 0.8701,
+      "step": 11122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.798639519677259e-07,
+      "loss": 0.7987,
+      "step": 11123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7934604789359044e-07,
+      "loss": 0.8411,
+      "step": 11124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7882888377998787e-07,
+      "loss": 0.7937,
+      "step": 11125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7831245966588583e-07,
+      "loss": 0.797,
+      "step": 11126
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7779677559019436e-07,
+      "loss": 0.868,
+      "step": 11127
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7728183159177126e-07,
+      "loss": 0.8498,
+      "step": 11128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7676762770941435e-07,
+      "loss": 0.8762,
+      "step": 11129
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7625416398186822e-07,
+      "loss": 0.8105,
+      "step": 11130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7574144044782083e-07,
+      "loss": 0.8693,
+      "step": 11131
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7522945714590566e-07,
+      "loss": 0.8847,
+      "step": 11132
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7471821411469747e-07,
+      "loss": 0.8599,
+      "step": 11133
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7420771139271765e-07,
+      "loss": 0.8666,
+      "step": 11134
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7369794901843207e-07,
+      "loss": 0.8431,
+      "step": 11135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7318892703024893e-07,
+      "loss": 0.8328,
+      "step": 11136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7268064546652308e-07,
+      "loss": 0.8302,
+      "step": 11137
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7217310436554945e-07,
+      "loss": 0.8277,
+      "step": 11138
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7166630376557414e-07,
+      "loss": 0.8512,
+      "step": 11139
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.711602437047788e-07,
+      "loss": 0.8782,
+      "step": 11140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7065492422129737e-07,
+      "loss": 0.8394,
+      "step": 11141
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7015034535320163e-07,
+      "loss": 0.8369,
+      "step": 11142
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6964650713851228e-07,
+      "loss": 0.8427,
+      "step": 11143
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6914340961519116e-07,
+      "loss": 0.8821,
+      "step": 11144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6864105282114573e-07,
+      "loss": 0.4836,
+      "step": 11145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6813943679422684e-07,
+      "loss": 0.4354,
+      "step": 11146
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.676385615722298e-07,
+      "loss": 0.8495,
+      "step": 11147
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.671384271928955e-07,
+      "loss": 0.8855,
+      "step": 11148
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.666390336939061e-07,
+      "loss": 0.8367,
+      "step": 11149
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6614038111289034e-07,
+      "loss": 0.834,
+      "step": 11150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6564246948742035e-07,
+      "loss": 0.8696,
+      "step": 11151
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6514529885501397e-07,
+      "loss": 0.4987,
+      "step": 11152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6464886925312895e-07,
+      "loss": 0.8305,
+      "step": 11153
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6415318071917207e-07,
+      "loss": 0.8688,
+      "step": 11154
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6365823329049124e-07,
+      "loss": 0.8509,
+      "step": 11155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6316402700437883e-07,
+      "loss": 0.8016,
+      "step": 11156
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6267056189807395e-07,
+      "loss": 0.8392,
+      "step": 11157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6217783800875576e-07,
+      "loss": 0.8344,
+      "step": 11158
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.616858553735512e-07,
+      "loss": 0.8508,
+      "step": 11159
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.611946140295295e-07,
+      "loss": 0.8163,
+      "step": 11160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6070411401370335e-07,
+      "loss": 0.8449,
+      "step": 11161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.60214355363032e-07,
+      "loss": 0.8635,
+      "step": 11162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5972533811441704e-07,
+      "loss": 0.5477,
+      "step": 11163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.592370623047046e-07,
+      "loss": 0.8633,
+      "step": 11164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5874952797068522e-07,
+      "loss": 0.8716,
+      "step": 11165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5826273514909284e-07,
+      "loss": 0.8521,
+      "step": 11166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5777668387660706e-07,
+      "loss": 0.8344,
+      "step": 11167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5729137418984852e-07,
+      "loss": 0.8673,
+      "step": 11168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.568068061253869e-07,
+      "loss": 0.8294,
+      "step": 11169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5632297971972966e-07,
+      "loss": 0.5185,
+      "step": 11170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5583989500933538e-07,
+      "loss": 0.8343,
+      "step": 11171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5535755203060054e-07,
+      "loss": 0.8769,
+      "step": 11172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.548759508198694e-07,
+      "loss": 0.8341,
+      "step": 11173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5439509141343068e-07,
+      "loss": 0.8412,
+      "step": 11174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5391497384751318e-07,
+      "loss": 0.8648,
+      "step": 11175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5343559815829468e-07,
+      "loss": 0.8352,
+      "step": 11176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5295696438189511e-07,
+      "loss": 0.8765,
+      "step": 11177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5247907255437456e-07,
+      "loss": 0.8633,
+      "step": 11178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.520019227117464e-07,
+      "loss": 0.8704,
+      "step": 11179
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5152551488995858e-07,
+      "loss": 0.8567,
+      "step": 11180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5104984912490906e-07,
+      "loss": 0.8689,
+      "step": 11181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.505749254524358e-07,
+      "loss": 0.8686,
+      "step": 11182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5010074390832684e-07,
+      "loss": 0.7824,
+      "step": 11183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4962730452830698e-07,
+      "loss": 0.8632,
+      "step": 11184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.49154607348051e-07,
+      "loss": 0.8515,
+      "step": 11185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.486826524031737e-07,
+      "loss": 0.855,
+      "step": 11186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.482114397292367e-07,
+      "loss": 0.8467,
+      "step": 11187
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4774096936174376e-07,
+      "loss": 0.8814,
+      "step": 11188
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4727124133614433e-07,
+      "loss": 0.8558,
+      "step": 11189
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4680225568783124e-07,
+      "loss": 0.8548,
+      "step": 11190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4633401245214064e-07,
+      "loss": 0.8674,
+      "step": 11191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4586651166435427e-07,
+      "loss": 0.8478,
+      "step": 11192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4539975335969735e-07,
+      "loss": 0.8077,
+      "step": 11193
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.449337375733373e-07,
+      "loss": 0.8362,
+      "step": 11194
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4446846434038708e-07,
+      "loss": 0.8166,
+      "step": 11195
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4400393369590649e-07,
+      "loss": 0.8776,
+      "step": 11196
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4354014567489528e-07,
+      "loss": 0.882,
+      "step": 11197
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.430771003122977e-07,
+      "loss": 0.8738,
+      "step": 11198
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.426147976430037e-07,
+      "loss": 0.8469,
+      "step": 11199
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4215323770184642e-07,
+      "loss": 0.8392,
+      "step": 11200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4169242052360367e-07,
+      "loss": 0.8407,
+      "step": 11201
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4123234614299651e-07,
+      "loss": 0.7899,
+      "step": 11202
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4077301459469062e-07,
+      "loss": 0.8648,
+      "step": 11203
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.403144259132949e-07,
+      "loss": 0.8464,
+      "step": 11204
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.398565801333629e-07,
+      "loss": 0.9001,
+      "step": 11205
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.393994772893925e-07,
+      "loss": 0.8641,
+      "step": 11206
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.389431174158251e-07,
+      "loss": 0.8492,
+      "step": 11207
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.384875005470454e-07,
+      "loss": 0.869,
+      "step": 11208
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.380326267173848e-07,
+      "loss": 0.8483,
+      "step": 11209
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3757849596111482e-07,
+      "loss": 0.8869,
+      "step": 11210
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3712510831245362e-07,
+      "loss": 0.9066,
+      "step": 11211
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3667246380556386e-07,
+      "loss": 0.8501,
+      "step": 11212
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.362205624745494e-07,
+      "loss": 0.8384,
+      "step": 11213
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3576940435346075e-07,
+      "loss": 0.9209,
+      "step": 11214
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3531898947629296e-07,
+      "loss": 0.8562,
+      "step": 11215
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3486931787697998e-07,
+      "loss": 0.8708,
+      "step": 11216
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3442038958940694e-07,
+      "loss": 0.8778,
+      "step": 11217
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.339722046473979e-07,
+      "loss": 0.8034,
+      "step": 11218
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.335247630847225e-07,
+      "loss": 0.8493,
+      "step": 11219
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.330780649350938e-07,
+      "loss": 0.859,
+      "step": 11220
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3263211023217038e-07,
+      "loss": 0.8708,
+      "step": 11221
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3218689900955207e-07,
+      "loss": 0.8756,
+      "step": 11222
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3174243130078645e-07,
+      "loss": 0.7777,
+      "step": 11223
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.312987071393612e-07,
+      "loss": 0.4395,
+      "step": 11224
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3085572655871183e-07,
+      "loss": 0.8587,
+      "step": 11225
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3041348959221377e-07,
+      "loss": 0.8402,
+      "step": 11226
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2997199627319047e-07,
+      "loss": 0.8358,
+      "step": 11227
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2953124663490412e-07,
+      "loss": 0.8686,
+      "step": 11228
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2909124071056712e-07,
+      "loss": 0.8921,
+      "step": 11229
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2865197853333179e-07,
+      "loss": 0.8307,
+      "step": 11230
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.282134601362939e-07,
+      "loss": 0.8396,
+      "step": 11231
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2777568555249697e-07,
+      "loss": 0.8445,
+      "step": 11232
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.273386548149247e-07,
+      "loss": 0.8433,
+      "step": 11233
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2690236795650734e-07,
+      "loss": 0.8411,
+      "step": 11234
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2646682501011642e-07,
+      "loss": 0.903,
+      "step": 11235
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.260320260085701e-07,
+      "loss": 0.8286,
+      "step": 11236
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.255979709846289e-07,
+      "loss": 0.8278,
+      "step": 11237
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.251646599709988e-07,
+      "loss": 0.9126,
+      "step": 11238
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2473209300032706e-07,
+      "loss": 0.8966,
+      "step": 11239
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2430027010520652e-07,
+      "loss": 0.8269,
+      "step": 11240
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2386919131817555e-07,
+      "loss": 0.8339,
+      "step": 11241
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2343885667171373e-07,
+      "loss": 0.8297,
+      "step": 11242
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2300926619824515e-07,
+      "loss": 0.829,
+      "step": 11243
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2258041993013947e-07,
+      "loss": 0.8214,
+      "step": 11244
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.221523178997075e-07,
+      "loss": 0.8417,
+      "step": 11245
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2172496013920788e-07,
+      "loss": 0.831,
+      "step": 11246
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2129834668083928e-07,
+      "loss": 0.8469,
+      "step": 11247
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2087247755674603e-07,
+      "loss": 0.8304,
+      "step": 11248
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2044735279901687e-07,
+      "loss": 0.8552,
+      "step": 11249
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2002297243968397e-07,
+      "loss": 0.9039,
+      "step": 11250
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1959933651072065e-07,
+      "loss": 0.8443,
+      "step": 11251
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1917644504405023e-07,
+      "loss": 0.8759,
+      "step": 11252
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1875429807153616e-07,
+      "loss": 0.8621,
+      "step": 11253
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1833289562498406e-07,
+      "loss": 0.8153,
+      "step": 11254
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1791223773614635e-07,
+      "loss": 0.8607,
+      "step": 11255
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1749232443671877e-07,
+      "loss": 0.8372,
+      "step": 11256
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1707315575834044e-07,
+      "loss": 0.8927,
+      "step": 11257
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1665473173259612e-07,
+      "loss": 0.817,
+      "step": 11258
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1623705239101058e-07,
+      "loss": 0.5552,
+      "step": 11259
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.158201177650553e-07,
+      "loss": 0.8052,
+      "step": 11260
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.154039278861474e-07,
+      "loss": 0.8528,
+      "step": 11261
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.149884827856429e-07,
+      "loss": 0.8912,
+      "step": 11262
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1457378249484674e-07,
+      "loss": 0.832,
+      "step": 11263
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1415982704500284e-07,
+      "loss": 0.8355,
+      "step": 11264
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1374661646730512e-07,
+      "loss": 0.8557,
+      "step": 11265
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1333415079288424e-07,
+      "loss": 0.8668,
+      "step": 11266
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.12922430052822e-07,
+      "loss": 0.8607,
+      "step": 11267
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1251145427813804e-07,
+      "loss": 0.8749,
+      "step": 11268
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1210122349979979e-07,
+      "loss": 0.8615,
+      "step": 11269
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1169173774871478e-07,
+      "loss": 0.9053,
+      "step": 11270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1128299705573942e-07,
+      "loss": 0.82,
+      "step": 11271
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1087500145166908e-07,
+      "loss": 0.8183,
+      "step": 11272
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1046775096724693e-07,
+      "loss": 0.8772,
+      "step": 11273
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1006124563315624e-07,
+      "loss": 0.5173,
+      "step": 11274
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0965548548002802e-07,
+      "loss": 0.8954,
+      "step": 11275
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0925047053843452e-07,
+      "loss": 0.8413,
+      "step": 11276
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0884620083889241e-07,
+      "loss": 0.8694,
+      "step": 11277
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.084426764118629e-07,
+      "loss": 0.8386,
+      "step": 11278
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0803989728774944e-07,
+      "loss": 0.8299,
+      "step": 11279
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0763786349690109e-07,
+      "loss": 0.8986,
+      "step": 11280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0723657506961027e-07,
+      "loss": 0.8296,
+      "step": 11281
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0683603203611281e-07,
+      "loss": 0.8343,
+      "step": 11282
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.064362344265879e-07,
+      "loss": 0.8579,
+      "step": 11283
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0603718227116034e-07,
+      "loss": 0.8688,
+      "step": 11284
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0563887559989606e-07,
+      "loss": 0.8552,
+      "step": 11285
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0524131444280882e-07,
+      "loss": 0.828,
+      "step": 11286
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0484449882985138e-07,
+      "loss": 0.8793,
+      "step": 11287
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0444842879092531e-07,
+      "loss": 0.8461,
+      "step": 11288
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0405310435587013e-07,
+      "loss": 0.8171,
+      "step": 11289
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0365852555447642e-07,
+      "loss": 0.8233,
+      "step": 11290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0326469241647153e-07,
+      "loss": 0.5133,
+      "step": 11291
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.028716049715306e-07,
+      "loss": 0.8537,
+      "step": 11292
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0247926324927215e-07,
+      "loss": 0.8191,
+      "step": 11293
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0208766727925812e-07,
+      "loss": 0.907,
+      "step": 11294
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0169681709099377e-07,
+      "loss": 0.8008,
+      "step": 11295
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0130671271392889e-07,
+      "loss": 0.85,
+      "step": 11296
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0091735417745662e-07,
+      "loss": 0.8646,
+      "step": 11297
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0052874151091463e-07,
+      "loss": 0.8589,
+      "step": 11298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0014087474358392e-07,
+      "loss": 0.8569,
+      "step": 11299
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.975375390468889e-08,
+      "loss": 0.8615,
+      "step": 11300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.936737902339844e-08,
+      "loss": 0.8587,
+      "step": 11301
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.89817501288226e-08,
+      "loss": 0.8513,
+      "step": 11302
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.85968672500215e-08,
+      "loss": 0.845,
+      "step": 11303
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.821273041599189e-08,
+      "loss": 0.7813,
+      "step": 11304
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.782933965567953e-08,
+      "loss": 0.8984,
+      "step": 11305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.744669499797021e-08,
+      "loss": 0.8592,
+      "step": 11306
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.706479647169643e-08,
+      "loss": 0.8386,
+      "step": 11307
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.66836441056318e-08,
+      "loss": 0.7986,
+      "step": 11308
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.630323792849672e-08,
+      "loss": 0.822,
+      "step": 11309
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.592357796895268e-08,
+      "loss": 0.7972,
+      "step": 11310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.554466425560793e-08,
+      "loss": 0.8438,
+      "step": 11311
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.516649681700962e-08,
+      "loss": 0.8664,
+      "step": 11312
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.478907568165385e-08,
+      "loss": 0.8442,
+      "step": 11313
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.441240087797787e-08,
+      "loss": 0.8369,
+      "step": 11314
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.403647243436342e-08,
+      "loss": 0.8327,
+      "step": 11315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.366129037913563e-08,
+      "loss": 0.8294,
+      "step": 11316
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.328685474056187e-08,
+      "loss": 0.5651,
+      "step": 11317
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.291316554685737e-08,
+      "loss": 0.8129,
+      "step": 11318
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.254022282617736e-08,
+      "loss": 0.8473,
+      "step": 11319
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.216802660662161e-08,
+      "loss": 0.828,
+      "step": 11320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.179657691623544e-08,
+      "loss": 0.9253,
+      "step": 11321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.142587378300539e-08,
+      "loss": 0.8329,
+      "step": 11322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.105591723486352e-08,
+      "loss": 0.878,
+      "step": 11323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.068670729968532e-08,
+      "loss": 0.8634,
+      "step": 11324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.031824400528854e-08,
+      "loss": 0.9195,
+      "step": 11325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.995052737943766e-08,
+      "loss": 0.8441,
+      "step": 11326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.958355744983937e-08,
+      "loss": 0.8481,
+      "step": 11327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.921733424414158e-08,
+      "loss": 0.8533,
+      "step": 11328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.885185778993999e-08,
+      "loss": 0.8345,
+      "step": 11329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.848712811477145e-08,
+      "loss": 0.8429,
+      "step": 11330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.812314524611843e-08,
+      "loss": 0.8889,
+      "step": 11331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.775990921140565e-08,
+      "loss": 0.8218,
+      "step": 11332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.739742003800233e-08,
+      "loss": 0.8816,
+      "step": 11333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.703567775321998e-08,
+      "loss": 0.8428,
+      "step": 11334
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.667468238431453e-08,
+      "loss": 0.7931,
+      "step": 11335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.63144339584887e-08,
+      "loss": 0.8055,
+      "step": 11336
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.595493250288301e-08,
+      "loss": 0.8602,
+      "step": 11337
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.559617804458798e-08,
+      "loss": 0.8189,
+      "step": 11338
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.5238170610632e-08,
+      "loss": 0.8469,
+      "step": 11339
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.488091022799239e-08,
+      "loss": 0.4129,
+      "step": 11340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.452439692358649e-08,
+      "loss": 0.8223,
+      "step": 11341
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.416863072427728e-08,
+      "loss": 0.8279,
+      "step": 11342
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.381361165687108e-08,
+      "loss": 0.8458,
+      "step": 11343
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.345933974811537e-08,
+      "loss": 0.865,
+      "step": 11344
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.310581502470661e-08,
+      "loss": 0.8829,
+      "step": 11345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.2753037513279e-08,
+      "loss": 0.8291,
+      "step": 11346
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.240100724041689e-08,
+      "loss": 0.8192,
+      "step": 11347
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.204972423264124e-08,
+      "loss": 0.8226,
+      "step": 11348
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.169918851642311e-08,
+      "loss": 0.8666,
+      "step": 11349
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.13494001181725e-08,
+      "loss": 0.8567,
+      "step": 11350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.100035906424719e-08,
+      "loss": 0.8215,
+      "step": 11351
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.065206538094395e-08,
+      "loss": 0.8127,
+      "step": 11352
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.030451909450842e-08,
+      "loss": 0.8473,
+      "step": 11353
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.995772023112525e-08,
+      "loss": 0.8393,
+      "step": 11354
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.961166881692573e-08,
+      "loss": 0.8695,
+      "step": 11355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.926636487798345e-08,
+      "loss": 0.8188,
+      "step": 11356
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.89218084403165e-08,
+      "loss": 0.8323,
+      "step": 11357
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.857799952988632e-08,
+      "loss": 0.8955,
+      "step": 11358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.823493817259776e-08,
+      "loss": 0.5064,
+      "step": 11359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.789262439430012e-08,
+      "loss": 0.7701,
+      "step": 11360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.755105822078502e-08,
+      "loss": 0.8515,
+      "step": 11361
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.721023967778962e-08,
+      "loss": 0.8516,
+      "step": 11362
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.68701687909923e-08,
+      "loss": 0.8486,
+      "step": 11363
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.653084558601808e-08,
+      "loss": 0.8537,
+      "step": 11364
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.619227008843322e-08,
+      "loss": 0.8588,
+      "step": 11365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.58544423237484e-08,
+      "loss": 0.8715,
+      "step": 11366
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.55173623174188e-08,
+      "loss": 0.8226,
+      "step": 11367
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.518103009484079e-08,
+      "loss": 0.9079,
+      "step": 11368
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.484544568135854e-08,
+      "loss": 0.819,
+      "step": 11369
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.451060910225627e-08,
+      "loss": 0.9012,
+      "step": 11370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.417652038276157e-08,
+      "loss": 0.8822,
+      "step": 11371
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.384317954804987e-08,
+      "loss": 0.8582,
+      "step": 11372
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.351058662323551e-08,
+      "loss": 0.868,
+      "step": 11373
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.317874163338068e-08,
+      "loss": 0.8589,
+      "step": 11374
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.284764460348647e-08,
+      "loss": 0.8405,
+      "step": 11375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.251729555850184e-08,
+      "loss": 0.8489,
+      "step": 11376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.218769452331576e-08,
+      "loss": 0.8608,
+      "step": 11377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.1858841522765e-08,
+      "loss": 0.8921,
+      "step": 11378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.153073658162646e-08,
+      "loss": 0.8672,
+      "step": 11379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.120337972462365e-08,
+      "loss": 0.8269,
+      "step": 11380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.08767709764202e-08,
+      "loss": 0.865,
+      "step": 11381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.055091036162642e-08,
+      "loss": 0.8701,
+      "step": 11382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.022579790479378e-08,
+      "loss": 0.8879,
+      "step": 11383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.990143363042045e-08,
+      "loss": 0.8707,
+      "step": 11384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.957781756294469e-08,
+      "loss": 0.8806,
+      "step": 11385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.925494972675029e-08,
+      "loss": 0.8713,
+      "step": 11386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.89328301461667e-08,
+      "loss": 0.8558,
+      "step": 11387
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.861145884546117e-08,
+      "loss": 0.871,
+      "step": 11388
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.82908358488521e-08,
+      "loss": 0.8146,
+      "step": 11389
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.797096118049462e-08,
+      "loss": 0.8338,
+      "step": 11390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.765183486449167e-08,
+      "loss": 0.8502,
+      "step": 11391
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.733345692488736e-08,
+      "loss": 0.8292,
+      "step": 11392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.701582738567359e-08,
+      "loss": 0.839,
+      "step": 11393
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.6698946270779e-08,
+      "loss": 0.8449,
+      "step": 11394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.638281360408339e-08,
+      "loss": 0.7921,
+      "step": 11395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.606742940940436e-08,
+      "loss": 0.8233,
+      "step": 11396
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.575279371050513e-08,
+      "loss": 0.8688,
+      "step": 11397
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.543890653109341e-08,
+      "loss": 0.8451,
+      "step": 11398
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.512576789482028e-08,
+      "loss": 0.5215,
+      "step": 11399
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.481337782527908e-08,
+      "loss": 0.8428,
+      "step": 11400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.450173634600876e-08,
+      "loss": 0.8326,
+      "step": 11401
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.419084348048832e-08,
+      "loss": 0.8625,
+      "step": 11402
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.388069925214458e-08,
+      "loss": 0.8786,
+      "step": 11403
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.35713036843455e-08,
+      "loss": 0.8855,
+      "step": 11404
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.326265680040467e-08,
+      "loss": 0.8462,
+      "step": 11405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.29547586235757e-08,
+      "loss": 0.8587,
+      "step": 11406
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.264760917705782e-08,
+      "loss": 0.8899,
+      "step": 11407
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.234120848399583e-08,
+      "loss": 0.8263,
+      "step": 11408
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.203555656747574e-08,
+      "loss": 0.8516,
+      "step": 11409
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.173065345052687e-08,
+      "loss": 0.8727,
+      "step": 11410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.142649915612308e-08,
+      "loss": 0.8193,
+      "step": 11411
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.112309370718162e-08,
+      "loss": 0.8051,
+      "step": 11412
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.082043712656305e-08,
+      "loss": 0.8198,
+      "step": 11413
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.051852943707248e-08,
+      "loss": 0.8101,
+      "step": 11414
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.021737066145727e-08,
+      "loss": 0.8355,
+      "step": 11415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.991696082240817e-08,
+      "loss": 0.8408,
+      "step": 11416
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.96172999425626e-08,
+      "loss": 0.8544,
+      "step": 11417
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9318388044498075e-08,
+      "loss": 0.8475,
+      "step": 11418
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9020225150735466e-08,
+      "loss": 0.9159,
+      "step": 11419
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.872281128374235e-08,
+      "loss": 0.9059,
+      "step": 11420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.842614646592637e-08,
+      "loss": 0.8317,
+      "step": 11421
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.813023071964296e-08,
+      "loss": 0.9192,
+      "step": 11422
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.783506406718653e-08,
+      "loss": 0.8624,
+      "step": 11423
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.754064653079816e-08,
+      "loss": 0.8452,
+      "step": 11424
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7246978132659e-08,
+      "loss": 0.8102,
+      "step": 11425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.695405889490024e-08,
+      "loss": 0.8712,
+      "step": 11426
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6661888839590895e-08,
+      "loss": 0.8138,
+      "step": 11427
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.637046798874335e-08,
+      "loss": 0.8216,
+      "step": 11428
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6079796364317815e-08,
+      "loss": 0.8695,
+      "step": 11429
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.578987398821345e-08,
+      "loss": 0.8462,
+      "step": 11430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.550070088227721e-08,
+      "loss": 0.8959,
+      "step": 11431
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5212277068297237e-08,
+      "loss": 0.845,
+      "step": 11432
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4924602568005024e-08,
+      "loss": 0.8059,
+      "step": 11433
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4924602568005024e-08,
+      "loss": 0.8484,
+      "step": 11434
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4637677403074355e-08,
+      "loss": 0.8415,
+      "step": 11435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.435150159512792e-08,
+      "loss": 0.8895,
+      "step": 11436
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.406607516572515e-08,
+      "loss": 0.822,
+      "step": 11437
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3781398136374394e-08,
+      "loss": 0.8702,
+      "step": 11438
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.349747052852405e-08,
+      "loss": 0.8858,
+      "step": 11439
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3214292363568123e-08,
+      "loss": 0.8752,
+      "step": 11440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2931863662841755e-08,
+      "loss": 0.8486,
+      "step": 11441
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.265018444762682e-08,
+      "loss": 0.8756,
+      "step": 11442
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.236925473914633e-08,
+      "loss": 0.8408,
+      "step": 11443
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.208907455856782e-08,
+      "loss": 0.8405,
+      "step": 11444
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.180964392700216e-08,
+      "loss": 0.9208,
+      "step": 11445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.153096286550474e-08,
+      "loss": 0.8576,
+      "step": 11446
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1253031395070985e-08,
+      "loss": 0.8504,
+      "step": 11447
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.097584953664525e-08,
+      "loss": 0.8463,
+      "step": 11448
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.069941731110972e-08,
+      "loss": 0.8445,
+      "step": 11449
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.04237347392944e-08,
+      "loss": 0.8911,
+      "step": 11450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0148801841971575e-08,
+      "loss": 0.8099,
+      "step": 11451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9874618639854656e-08,
+      "loss": 0.8382,
+      "step": 11452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.96011851536049e-08,
+      "loss": 0.86,
+      "step": 11453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9328501403824725e-08,
+      "loss": 0.8739,
+      "step": 11454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.905656741105769e-08,
+      "loss": 0.8543,
+      "step": 11455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.878538319579629e-08,
+      "loss": 0.4466,
+      "step": 11456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.851494877847196e-08,
+      "loss": 0.8869,
+      "step": 11457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.824526417946063e-08,
+      "loss": 0.8643,
+      "step": 11458
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7976329419084924e-08,
+      "loss": 0.8322,
+      "step": 11459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.77081445176053e-08,
+      "loss": 0.8448,
+      "step": 11460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7440709495231164e-08,
+      "loss": 0.8427,
+      "step": 11461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.717402437211083e-08,
+      "loss": 0.8826,
+      "step": 11462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6908089168340444e-08,
+      "loss": 0.8401,
+      "step": 11463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.664290390395509e-08,
+      "loss": 0.8587,
+      "step": 11464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6378468598938794e-08,
+      "loss": 0.8156,
+      "step": 11465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6114783273213395e-08,
+      "loss": 0.8353,
+      "step": 11466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5851847946649653e-08,
+      "loss": 0.857,
+      "step": 11467
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5589662639056175e-08,
+      "loss": 0.8502,
+      "step": 11468
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.532822737018938e-08,
+      "loss": 0.8272,
+      "step": 11469
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5067542159747955e-08,
+      "loss": 0.8541,
+      "step": 11470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.480760702737286e-08,
+      "loss": 0.8601,
+      "step": 11471
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.454842199265064e-08,
+      "loss": 0.8676,
+      "step": 11472
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4289987075109006e-08,
+      "loss": 0.8494,
+      "step": 11473
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.40323022942224e-08,
+      "loss": 0.8222,
+      "step": 11474
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.377536766940416e-08,
+      "loss": 0.8608,
+      "step": 11475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.351918322001547e-08,
+      "loss": 0.8562,
+      "step": 11476
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.326374896535757e-08,
+      "loss": 0.8475,
+      "step": 11477
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.300906492467838e-08,
+      "loss": 0.8653,
+      "step": 11478
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.275513111716589e-08,
+      "loss": 0.8654,
+      "step": 11479
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2501947561955914e-08,
+      "loss": 0.8601,
+      "step": 11480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2249514278123184e-08,
+      "loss": 0.8137,
+      "step": 11481
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.199783128468804e-08,
+      "loss": 0.8498,
+      "step": 11482
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.174689860061532e-08,
+      "loss": 0.8435,
+      "step": 11483
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1496716244811e-08,
+      "loss": 0.8298,
+      "step": 11484
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1247284236125565e-08,
+      "loss": 0.8537,
+      "step": 11485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.099860259335287e-08,
+      "loss": 0.8412,
+      "step": 11486
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0750671335232363e-08,
+      "loss": 0.8441,
+      "step": 11487
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.050349048044244e-08,
+      "loss": 0.8831,
+      "step": 11488
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.025706004760932e-08,
+      "loss": 0.8968,
+      "step": 11489
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.001138005530036e-08,
+      "loss": 0.512,
+      "step": 11490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9766450522027436e-08,
+      "loss": 0.8304,
+      "step": 11491
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.952227146624465e-08,
+      "loss": 0.8686,
+      "step": 11492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.927884290634953e-08,
+      "loss": 0.517,
+      "step": 11493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.903616486068518e-08,
+      "loss": 0.8465,
+      "step": 11494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.879423734753585e-08,
+      "loss": 0.8572,
+      "step": 11495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.85530603851314e-08,
+      "loss": 0.8405,
+      "step": 11496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8312633991642865e-08,
+      "loss": 0.8679,
+      "step": 11497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8072958185184635e-08,
+      "loss": 0.8652,
+      "step": 11498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7834032983817826e-08,
+      "loss": 0.8405,
+      "step": 11499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.759585840554358e-08,
+      "loss": 0.8256,
+      "step": 11500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.735843446830867e-08,
+      "loss": 0.8126,
+      "step": 11501
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.712176119000099e-08,
+      "loss": 0.514,
+      "step": 11502
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.688583858845518e-08,
+      "loss": 0.8539,
+      "step": 11503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.665066668144479e-08,
+      "loss": 0.8038,
+      "step": 11504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6416245486692315e-08,
+      "loss": 0.8936,
+      "step": 11505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.618257502185696e-08,
+      "loss": 0.8668,
+      "step": 11506
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5949655304550193e-08,
+      "loss": 0.8149,
+      "step": 11507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.571748635231687e-08,
+      "loss": 0.8423,
+      "step": 11508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.54860681826541e-08,
+      "loss": 0.8535,
+      "step": 11509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.525540081299683e-08,
+      "loss": 0.8331,
+      "step": 11510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.502548426072561e-08,
+      "loss": 0.8562,
+      "step": 11511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.479631854316434e-08,
+      "loss": 0.8599,
+      "step": 11512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.456790367757923e-08,
+      "loss": 0.8682,
+      "step": 11513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.434023968118205e-08,
+      "loss": 0.8773,
+      "step": 11514
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.411332657112465e-08,
+      "loss": 0.8727,
+      "step": 11515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3887164364506676e-08,
+      "loss": 0.8269,
+      "step": 11516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.366175307836672e-08,
+      "loss": 0.8027,
+      "step": 11517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.34370927296912e-08,
+      "loss": 0.8173,
+      "step": 11518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.321318333540546e-08,
+      "loss": 0.8718,
+      "step": 11519
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.299002491238268e-08,
+      "loss": 0.8528,
+      "step": 11520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.276761747743495e-08,
+      "loss": 0.4709,
+      "step": 11521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.254596104732222e-08,
+      "loss": 0.8551,
+      "step": 11522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.232505563874444e-08,
+      "loss": 0.8525,
+      "step": 11523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2104901268346086e-08,
+      "loss": 0.8439,
+      "step": 11524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.188549795271612e-08,
+      "loss": 0.8208,
+      "step": 11525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.166684570838463e-08,
+      "loss": 0.8174,
+      "step": 11526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1448944551828454e-08,
+      "loss": 0.8099,
+      "step": 11527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.123179449946445e-08,
+      "loss": 0.8585,
+      "step": 11528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.101539556765509e-08,
+      "loss": 0.8298,
+      "step": 11529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0799747772704006e-08,
+      "loss": 0.8251,
+      "step": 11530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0584851130861516e-08,
+      "loss": 0.8476,
+      "step": 11531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0370705658319125e-08,
+      "loss": 0.8375,
+      "step": 11532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.015731137121169e-08,
+      "loss": 0.5259,
+      "step": 11533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9944668285617486e-08,
+      "loss": 0.9052,
+      "step": 11534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.973277641755923e-08,
+      "loss": 0.6042,
+      "step": 11535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9521635783001932e-08,
+      "loss": 0.8347,
+      "step": 11536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9311246397855097e-08,
+      "loss": 0.7873,
+      "step": 11537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9101608277970484e-08,
+      "loss": 0.8414,
+      "step": 11538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8892721439144344e-08,
+      "loss": 0.8584,
+      "step": 11539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8684585897116313e-08,
+      "loss": 0.4616,
+      "step": 11540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.847720166756718e-08,
+      "loss": 0.8012,
+      "step": 11541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.827056876612333e-08,
+      "loss": 0.815,
+      "step": 11542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8064687208354534e-08,
+      "loss": 0.8368,
+      "step": 11543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.785955700977394e-08,
+      "loss": 0.8345,
+      "step": 11544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.765517818583585e-08,
+      "loss": 0.8573,
+      "step": 11545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7451550751941282e-08,
+      "loss": 0.8207,
+      "step": 11546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.72486747234324e-08,
+      "loss": 0.8289,
+      "step": 11547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.704655011559476e-08,
+      "loss": 0.8786,
+      "step": 11548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.684517694365951e-08,
+      "loss": 0.7992,
+      "step": 11549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.664455522279785e-08,
+      "loss": 0.8528,
+      "step": 11550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6444684968127687e-08,
+      "loss": 0.4615,
+      "step": 11551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6245566194706973e-08,
+      "loss": 0.8877,
+      "step": 11552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6047198917540372e-08,
+      "loss": 0.8762,
+      "step": 11553
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.584958315157371e-08,
+      "loss": 0.8824,
+      "step": 11554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5652718911696185e-08,
+      "loss": 0.8628,
+      "step": 11555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5456606212742596e-08,
+      "loss": 0.86,
+      "step": 11556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5261245069487793e-08,
+      "loss": 0.8219,
+      "step": 11557
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5066635496652225e-08,
+      "loss": 0.8198,
+      "step": 11558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4872777508899716e-08,
+      "loss": 0.8436,
+      "step": 11559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4679671120837466e-08,
+      "loss": 0.8249,
+      "step": 11560
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4487316347013845e-08,
+      "loss": 0.8054,
+      "step": 11561
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4295713201923922e-08,
+      "loss": 0.8566,
+      "step": 11562
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4104861700002813e-08,
+      "loss": 0.507,
+      "step": 11563
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3914761855632353e-08,
+      "loss": 0.8368,
+      "step": 11564
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3725413683135525e-08,
+      "loss": 0.8545,
+      "step": 11565
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3536817196778693e-08,
+      "loss": 0.8488,
+      "step": 11566
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3348972410772718e-08,
+      "loss": 0.8607,
+      "step": 11567
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.316187933927072e-08,
+      "loss": 0.9009,
+      "step": 11568
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2975537996370313e-08,
+      "loss": 0.8567,
+      "step": 11569
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.278994839611026e-08,
+      "loss": 0.845,
+      "step": 11570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2605110552477162e-08,
+      "loss": 0.8899,
+      "step": 11571
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2421024479395426e-08,
+      "loss": 0.848,
+      "step": 11572
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2237690190736183e-08,
+      "loss": 0.8476,
+      "step": 11573
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2055107700312827e-08,
+      "loss": 0.8539,
+      "step": 11574
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1873277021884353e-08,
+      "loss": 0.8966,
+      "step": 11575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.169219816914869e-08,
+      "loss": 0.8797,
+      "step": 11576
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1511871155751597e-08,
+      "loss": 0.8123,
+      "step": 11577
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1332295995279972e-08,
+      "loss": 0.8332,
+      "step": 11578
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1153472701263e-08,
+      "loss": 0.8501,
+      "step": 11579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0975401287175456e-08,
+      "loss": 0.8283,
+      "step": 11580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0798081766434385e-08,
+      "loss": 0.8413,
+      "step": 11581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0621514152401323e-08,
+      "loss": 0.8214,
+      "step": 11582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0445698458377848e-08,
+      "loss": 0.8519,
+      "step": 11583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0270634697614478e-08,
+      "loss": 0.8436,
+      "step": 11584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0096322883298435e-08,
+      "loss": 0.8715,
+      "step": 11585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9922763028565885e-08,
+      "loss": 0.8455,
+      "step": 11586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9749955146494137e-08,
+      "loss": 0.8506,
+      "step": 11587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9577899250101674e-08,
+      "loss": 0.8675,
+      "step": 11588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9406595352354784e-08,
+      "loss": 0.8606,
+      "step": 11589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9236043466159816e-08,
+      "loss": 0.8294,
+      "step": 11590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9066243604367595e-08,
+      "loss": 0.7841,
+      "step": 11591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8897195779771225e-08,
+      "loss": 0.866,
+      "step": 11592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8728900005109408e-08,
+      "loss": 0.8396,
+      "step": 11593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8561356293061995e-08,
+      "loss": 0.8414,
+      "step": 11594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8394564656253335e-08,
+      "loss": 0.8464,
+      "step": 11595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8228525107251148e-08,
+      "loss": 0.846,
+      "step": 11596
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8063237658564325e-08,
+      "loss": 0.8455,
+      "step": 11597
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7898702322648453e-08,
+      "loss": 0.8406,
+      "step": 11598
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7734919111900286e-08,
+      "loss": 0.8744,
+      "step": 11599
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7571888038661056e-08,
+      "loss": 0.8696,
+      "step": 11600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7409609115213166e-08,
+      "loss": 0.8885,
+      "step": 11601
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7248082353785723e-08,
+      "loss": 0.8091,
+      "step": 11602
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.708730776654899e-08,
+      "loss": 0.8345,
+      "step": 11603
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6927285365615497e-08,
+      "loss": 0.8107,
+      "step": 11604
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6768015163043383e-08,
+      "loss": 0.8284,
+      "step": 11605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6609497170834154e-08,
+      "loss": 0.501,
+      "step": 11606
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6451731400930483e-08,
+      "loss": 0.8353,
+      "step": 11607
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6294717865220634e-08,
+      "loss": 0.8561,
+      "step": 11608
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6138456575534034e-08,
+      "loss": 0.8423,
+      "step": 11609
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.598294754364571e-08,
+      "loss": 0.8071,
+      "step": 11610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5828190781271846e-08,
+      "loss": 0.878,
+      "step": 11611
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5674186300073113e-08,
+      "loss": 0.8659,
+      "step": 11612
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5520934111653563e-08,
+      "loss": 0.8152,
+      "step": 11613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5368434227560624e-08,
+      "loss": 0.7988,
+      "step": 11614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5216686659285106e-08,
+      "loss": 0.8341,
+      "step": 11615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5065691418260086e-08,
+      "loss": 0.8168,
+      "step": 11616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.491544851586202e-08,
+      "loss": 0.876,
+      "step": 11617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4765957963412957e-08,
+      "loss": 0.8687,
+      "step": 11618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4617219772175006e-08,
+      "loss": 0.9307,
+      "step": 11619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4469233953355866e-08,
+      "loss": 0.469,
+      "step": 11620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4322000518106616e-08,
+      "loss": 0.8781,
+      "step": 11621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4175519477520606e-08,
+      "loss": 0.9208,
+      "step": 11622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4029790842633451e-08,
+      "loss": 0.8551,
+      "step": 11623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3884814624427478e-08,
+      "loss": 0.5224,
+      "step": 11624
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3740590833823952e-08,
+      "loss": 0.8429,
+      "step": 11625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3597119481691956e-08,
+      "loss": 0.8348,
+      "step": 11626
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.345440057884062e-08,
+      "loss": 0.8037,
+      "step": 11627
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3312434136023567e-08,
+      "loss": 0.8878,
+      "step": 11628
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3171220163936682e-08,
+      "loss": 0.8447,
+      "step": 11629
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3030758673221456e-08,
+      "loss": 0.905,
+      "step": 11630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2891049674461643e-08,
+      "loss": 0.7921,
+      "step": 11631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2752093178182157e-08,
+      "loss": 0.9053,
+      "step": 11632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2613889194854623e-08,
+      "loss": 0.8601,
+      "step": 11633
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2476437734891822e-08,
+      "loss": 0.8457,
+      "step": 11634
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2339738808649915e-08,
+      "loss": 0.8336,
+      "step": 11635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.220379242642844e-08,
+      "loss": 0.859,
+      "step": 11636
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2068598598472536e-08,
+      "loss": 0.8581,
+      "step": 11637
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.193415733496628e-08,
+      "loss": 0.8697,
+      "step": 11638
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1800468646041563e-08,
+      "loss": 0.8498,
+      "step": 11639
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1667532541769222e-08,
+      "loss": 0.8074,
+      "step": 11640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1535349032167908e-08,
+      "loss": 0.8796,
+      "step": 11641
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1403918127196323e-08,
+      "loss": 0.8726,
+      "step": 11642
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1273239836757655e-08,
+      "loss": 0.8275,
+      "step": 11643
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1143314170698472e-08,
+      "loss": 0.8468,
+      "step": 11644
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.10141411388065e-08,
+      "loss": 0.821,
+      "step": 11645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0885720750817286e-08,
+      "loss": 0.8354,
+      "step": 11646
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0758053016405312e-08,
+      "loss": 0.8884,
+      "step": 11647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.063113794518955e-08,
+      "loss": 0.8392,
+      "step": 11648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0504975546733464e-08,
+      "loss": 0.8314,
+      "step": 11649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0379565830543892e-08,
+      "loss": 0.8647,
+      "step": 11650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0254908806068831e-08,
+      "loss": 0.8545,
+      "step": 11651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0131004482701878e-08,
+      "loss": 0.8613,
+      "step": 11652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.00078528697789e-08,
+      "loss": 0.862,
+      "step": 11653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.88545397657692e-09,
+      "loss": 0.8646,
+      "step": 11654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.76380781232078e-09,
+      "loss": 0.8221,
+      "step": 11655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.642914386175373e-09,
+      "loss": 0.8659,
+      "step": 11656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.522773707250077e-09,
+      "loss": 0.8603,
+      "step": 11657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.403385784596541e-09,
+      "loss": 0.8375,
+      "step": 11658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.284750627210904e-09,
+      "loss": 0.8619,
+      "step": 11659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.166868244031568e-09,
+      "loss": 0.8672,
+      "step": 11660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.04973864394032e-09,
+      "loss": 0.8577,
+      "step": 11661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.933361835763433e-09,
+      "loss": 0.8185,
+      "step": 11662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.817737828269446e-09,
+      "loss": 0.8173,
+      "step": 11663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.702866630170282e-09,
+      "loss": 0.8438,
+      "step": 11664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.588748250120126e-09,
+      "loss": 0.8574,
+      "step": 11665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.47538269671988e-09,
+      "loss": 0.8634,
+      "step": 11666
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.362769978508267e-09,
+      "loss": 0.862,
+      "step": 11667
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.250910103972942e-09,
+      "loss": 0.7889,
+      "step": 11668
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.139803081540499e-09,
+      "loss": 0.843,
+      "step": 11669
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.029448919584237e-09,
+      "loss": 0.8592,
+      "step": 11670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.91984762641751e-09,
+      "loss": 0.8767,
+      "step": 11671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.810999210299263e-09,
+      "loss": 0.9286,
+      "step": 11672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.702903679430718e-09,
+      "loss": 0.8425,
+      "step": 11673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.595561041956468e-09,
+      "loss": 0.891,
+      "step": 11674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.488971305965598e-09,
+      "loss": 0.8479,
+      "step": 11675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.3831344794872415e-09,
+      "loss": 0.8612,
+      "step": 11676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.2780505704983515e-09,
+      "loss": 0.8715,
+      "step": 11677
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.173719586914818e-09,
+      "loss": 0.8896,
+      "step": 11678
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.07014153659813e-09,
+      "loss": 0.495,
+      "step": 11679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.967316427353154e-09,
+      "loss": 0.843,
+      "step": 11680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.8652442669281394e-09,
+      "loss": 0.8453,
+      "step": 11681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.7639250630113785e-09,
+      "loss": 0.8766,
+      "step": 11682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.663358823240096e-09,
+      "loss": 0.8644,
+      "step": 11683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.563545555189343e-09,
+      "loss": 0.8534,
+      "step": 11684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.464485266380882e-09,
+      "loss": 0.8407,
+      "step": 11685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3661779642787415e-09,
+      "loss": 0.8624,
+      "step": 11686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.2686236562903294e-09,
+      "loss": 0.9097,
+      "step": 11687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.171822349765322e-09,
+      "loss": 0.8406,
+      "step": 11688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0757740519978844e-09,
+      "loss": 0.857,
+      "step": 11689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.980478770224452e-09,
+      "loss": 0.8625,
+      "step": 11690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.885936511625945e-09,
+      "loss": 0.8675,
+      "step": 11691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.7921472833255556e-09,
+      "loss": 0.8321,
+      "step": 11692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.699111092389853e-09,
+      "loss": 0.8174,
+      "step": 11693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6068279458298954e-09,
+      "loss": 0.8583,
+      "step": 11694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5152978505979e-09,
+      "loss": 0.8295,
+      "step": 11695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4245208135905725e-09,
+      "loss": 0.5239,
+      "step": 11696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.334496841647996e-09,
+      "loss": 0.8428,
+      "step": 11697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.245225941552523e-09,
+      "loss": 0.7988,
+      "step": 11698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.156708120032106e-09,
+      "loss": 0.8401,
+      "step": 11699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0689433837547434e-09,
+      "loss": 0.8703,
+      "step": 11700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.981931739334034e-09,
+      "loss": 0.8311,
+      "step": 11701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.895673193325845e-09,
+      "loss": 0.826,
+      "step": 11702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8101677522305325e-09,
+      "loss": 0.8837,
+      "step": 11703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7254154224885e-09,
+      "loss": 0.8563,
+      "step": 11704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.64141621048797e-09,
+      "loss": 0.8404,
+      "step": 11705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.558170122556105e-09,
+      "loss": 0.8198,
+      "step": 11706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.475677164966774e-09,
+      "loss": 0.8492,
+      "step": 11707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.393937343933896e-09,
+      "loss": 0.8615,
+      "step": 11708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.312950665618098e-09,
+      "loss": 0.8327,
+      "step": 11709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.232717136121167e-09,
+      "loss": 0.8942,
+      "step": 11710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.153236761488266e-09,
+      "loss": 0.8233,
+      "step": 11711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.074509547707939e-09,
+      "loss": 0.7938,
+      "step": 11712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.996535500712107e-09,
+      "loss": 0.8252,
+      "step": 11713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9193146263749595e-09,
+      "loss": 0.8371,
+      "step": 11714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.842846930517397e-09,
+      "loss": 0.8444,
+      "step": 11715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.767132418898145e-09,
+      "loss": 0.8263,
+      "step": 11716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.692171097223751e-09,
+      "loss": 0.8615,
+      "step": 11717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6179629711419195e-09,
+      "loss": 0.8102,
+      "step": 11718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.544508046244843e-09,
+      "loss": 0.8734,
+      "step": 11719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.471806328065874e-09,
+      "loss": 0.8175,
+      "step": 11720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.399857822083963e-09,
+      "loss": 0.8747,
+      "step": 11721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.328662533719218e-09,
+      "loss": 0.8579,
+      "step": 11722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2582204683362372e-09,
+      "loss": 0.879,
+      "step": 11723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1885316312441073e-09,
+      "loss": 0.8448,
+      "step": 11724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1195960276919624e-09,
+      "loss": 0.8503,
+      "step": 11725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0514136628745363e-09,
+      "loss": 0.8224,
+      "step": 11726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9839845419288303e-09,
+      "loss": 0.8095,
+      "step": 11727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9173086699363363e-09,
+      "loss": 0.8627,
+      "step": 11728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.851386051919702e-09,
+      "loss": 0.8182,
+      "step": 11729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7862166928471766e-09,
+      "loss": 0.8447,
+      "step": 11730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.721800597629276e-09,
+      "loss": 0.8516,
+      "step": 11731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6581377711176747e-09,
+      "loss": 0.8692,
+      "step": 11732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.595228218110757e-09,
+      "loss": 0.8481,
+      "step": 11733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5330719433491747e-09,
+      "loss": 0.8446,
+      "step": 11734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4716689515147386e-09,
+      "loss": 0.8571,
+      "step": 11735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.411019247234858e-09,
+      "loss": 0.835,
+      "step": 11736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.351122835078101e-09,
+      "loss": 0.5181,
+      "step": 11737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.291979719559745e-09,
+      "loss": 0.8246,
+      "step": 11738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.233589905134004e-09,
+      "loss": 0.8195,
+      "step": 11739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.175953396201802e-09,
+      "loss": 0.8523,
+      "step": 11740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1190701971052218e-09,
+      "loss": 0.8605,
+      "step": 11741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0629403121297244e-09,
+      "loss": 0.799,
+      "step": 11742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.007563745505259e-09,
+      "loss": 0.8626,
+      "step": 11743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.952940501405154e-09,
+      "loss": 0.7764,
+      "step": 11744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.899070583943896e-09,
+      "loss": 0.548,
+      "step": 11745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8459539971804608e-09,
+      "loss": 0.8884,
+      "step": 11746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7935907451172019e-09,
+      "loss": 0.8913,
+      "step": 11747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.741980831699852e-09,
+      "loss": 0.9177,
+      "step": 11748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6911242608164124e-09,
+      "loss": 0.8186,
+      "step": 11749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6410210362993729e-09,
+      "loss": 0.8555,
+      "step": 11750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5916711619246017e-09,
+      "loss": 0.8615,
+      "step": 11751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5430746414091259e-09,
+      "loss": 0.796,
+      "step": 11752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4952314784144606e-09,
+      "loss": 0.8581,
+      "step": 11753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4481416765466104e-09,
+      "loss": 0.8529,
+      "step": 11754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4018052393538483e-09,
+      "loss": 0.8725,
+      "step": 11755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3562221703267153e-09,
+      "loss": 0.8697,
+      "step": 11756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3113924728991311e-09,
+      "loss": 0.8635,
+      "step": 11757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2673161504495046e-09,
+      "loss": 0.7897,
+      "step": 11758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2239932062996229e-09,
+      "loss": 0.8308,
+      "step": 11759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1814236437135419e-09,
+      "loss": 0.7629,
+      "step": 11760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1396074658975854e-09,
+      "loss": 0.8646,
+      "step": 11761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0985446760036766e-09,
+      "loss": 0.8261,
+      "step": 11762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0582352771260074e-09,
+      "loss": 0.8009,
+      "step": 11763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0186792722999272e-09,
+      "loss": 0.8683,
+      "step": 11764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.798766645074953e-10,
+      "loss": 0.8721,
+      "step": 11765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.418274566719289e-10,
+      "loss": 0.5244,
+      "step": 11766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.045316516609337e-10,
+      "loss": 0.8628,
+      "step": 11767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.679892522833744e-10,
+      "loss": 0.862,
+      "step": 11768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.322002612937141e-10,
+      "loss": 0.8481,
+      "step": 11769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.971646813875744e-10,
+      "loss": 0.5243,
+      "step": 11770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.628825152050656e-10,
+      "loss": 0.8639,
+      "step": 11771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.293537653296768e-10,
+      "loss": 0.8705,
+      "step": 11772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.965784342871651e-10,
+      "loss": 0.829,
+      "step": 11773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.64556524547777e-10,
+      "loss": 0.7877,
+      "step": 11774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.332880385240269e-10,
+      "loss": 0.4558,
+      "step": 11775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.027729785718084e-10,
+      "loss": 0.8594,
+      "step": 11776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.730113469903931e-10,
+      "loss": 0.8567,
+      "step": 11777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.440031460213213e-10,
+      "loss": 0.8276,
+      "step": 11778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.157483778517325e-10,
+      "loss": 0.8089,
+      "step": 11779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.882470446099241e-10,
+      "loss": 0.8722,
+      "step": 11780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.614991483686826e-10,
+      "loss": 0.4883,
+      "step": 11781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3550469114195246e-10,
+      "loss": 0.8066,
+      "step": 11782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1026367488927745e-10,
+      "loss": 0.8706,
+      "step": 11783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.857761015124695e-10,
+      "loss": 0.8328,
+      "step": 11784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.620419728567193e-10,
+      "loss": 0.8462,
+      "step": 11785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.390612907094859e-10,
+      "loss": 0.8626,
+      "step": 11786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.168340568038275e-10,
+      "loss": 0.8498,
+      "step": 11787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.953602728128502e-10,
+      "loss": 0.8415,
+      "step": 11788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.746399403552591e-10,
+      "loss": 0.8683,
+      "step": 11789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5467306099313806e-10,
+      "loss": 0.8349,
+      "step": 11790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.35459636229729e-10,
+      "loss": 0.8427,
+      "step": 11791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1699966751387303e-10,
+      "loss": 0.8459,
+      "step": 11792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.992931562344591e-10,
+      "loss": 0.8061,
+      "step": 11793
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8234010372819576e-10,
+      "loss": 0.8596,
+      "step": 11794
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6614051127072929e-10,
+      "loss": 0.8369,
+      "step": 11795
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5069438008330494e-10,
+      "loss": 0.8767,
+      "step": 11796
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.360017113283263e-10,
+      "loss": 0.8346,
+      "step": 11797
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2206250611490612e-10,
+      "loss": 0.5977,
+      "step": 11798
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0887676549331539e-10,
+      "loss": 0.8623,
+      "step": 11799
+    },
+    {
+      "epoch": 1.0,
+      "step": 11799,
+      "total_flos": 8303741185359872.0,
+      "train_loss": 0.9039537689702795,
+      "train_runtime": 128891.5857,
+      "train_samples_per_second": 11.717,
+      "train_steps_per_second": 0.092
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 11799,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 8303741185359872.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}