diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31813 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9991508632889894,
+  "global_step": 5298,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2578616352201258e-07,
+      "loss": 1.8862,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5157232704402517e-07,
+      "loss": 1.853,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.773584905660378e-07,
+      "loss": 2.1465,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.031446540880503e-07,
+      "loss": 2.0024,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.28930817610063e-07,
+      "loss": 1.8237,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.547169811320755e-07,
+      "loss": 1.8203,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.805031446540881e-07,
+      "loss": 1.8716,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0062893081761007e-06,
+      "loss": 1.9595,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1320754716981133e-06,
+      "loss": 1.9653,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.257861635220126e-06,
+      "loss": 1.8979,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3836477987421384e-06,
+      "loss": 1.812,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.509433962264151e-06,
+      "loss": 1.7827,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6352201257861635e-06,
+      "loss": 1.687,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7610062893081762e-06,
+      "loss": 1.5278,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8867924528301889e-06,
+      "loss": 1.5571,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0125786163522013e-06,
+      "loss": 1.4976,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.138364779874214e-06,
+      "loss": 1.481,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2641509433962266e-06,
+      "loss": 1.3154,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.389937106918239e-06,
+      "loss": 1.3984,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.515723270440252e-06,
+      "loss": 1.3276,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6415094339622644e-06,
+      "loss": 1.3579,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.767295597484277e-06,
+      "loss": 1.3418,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8930817610062893e-06,
+      "loss": 1.3167,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.018867924528302e-06,
+      "loss": 1.3218,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1446540880503146e-06,
+      "loss": 1.271,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.270440251572327e-06,
+      "loss": 1.2656,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3962264150943395e-06,
+      "loss": 1.3062,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5220125786163524e-06,
+      "loss": 1.2935,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.647798742138365e-06,
+      "loss": 1.2661,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7735849056603777e-06,
+      "loss": 1.2305,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.89937106918239e-06,
+      "loss": 1.1592,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.025157232704403e-06,
+      "loss": 1.2334,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.150943396226416e-06,
+      "loss": 1.2222,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.276729559748428e-06,
+      "loss": 1.2437,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.402515723270441e-06,
+      "loss": 1.1804,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.528301886792453e-06,
+      "loss": 1.1802,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.654088050314466e-06,
+      "loss": 1.2244,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.779874213836478e-06,
+      "loss": 1.229,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.905660377358491e-06,
+      "loss": 1.2314,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.031446540880504e-06,
+      "loss": 1.166,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.157232704402516e-06,
+      "loss": 1.2266,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.283018867924529e-06,
+      "loss": 1.2236,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.408805031446541e-06,
+      "loss": 1.2402,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.534591194968554e-06,
+      "loss": 1.1133,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.660377358490566e-06,
+      "loss": 1.1836,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.786163522012579e-06,
+      "loss": 1.1909,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.911949685534591e-06,
+      "loss": 1.1611,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.037735849056604e-06,
+      "loss": 1.1621,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.163522012578617e-06,
+      "loss": 1.1111,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.289308176100629e-06,
+      "loss": 1.1233,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.415094339622642e-06,
+      "loss": 1.1487,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.540880503144654e-06,
+      "loss": 1.1489,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0837,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.792452830188679e-06,
+      "loss": 1.125,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9182389937106915e-06,
+      "loss": 1.1353,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.044025157232705e-06,
+      "loss": 1.1011,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.169811320754717e-06,
+      "loss": 1.1189,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.29559748427673e-06,
+      "loss": 1.123,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.421383647798742e-06,
+      "loss": 1.1191,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.5471698113207555e-06,
+      "loss": 1.0947,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.672955974842768e-06,
+      "loss": 1.0835,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.79874213836478e-06,
+      "loss": 1.1086,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.924528301886793e-06,
+      "loss": 1.1177,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.050314465408805e-06,
+      "loss": 1.1025,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.17610062893082e-06,
+      "loss": 1.1467,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.301886792452832e-06,
+      "loss": 1.0874,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.427672955974844e-06,
+      "loss": 1.1084,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.553459119496857e-06,
+      "loss": 1.1064,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.67924528301887e-06,
+      "loss": 1.1421,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.805031446540882e-06,
+      "loss": 1.1013,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.930817610062894e-06,
+      "loss": 1.1196,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.056603773584907e-06,
+      "loss": 1.1201,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.182389937106919e-06,
+      "loss": 1.103,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.308176100628931e-06,
+      "loss": 1.1255,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.433962264150944e-06,
+      "loss": 1.0977,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.559748427672956e-06,
+      "loss": 1.1033,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.685534591194969e-06,
+      "loss": 1.0649,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.811320754716981e-06,
+      "loss": 1.1179,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.937106918238994e-06,
+      "loss": 1.1189,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0062893081761008e-05,
+      "loss": 1.1184,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.018867924528302e-05,
+      "loss": 1.0903,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0314465408805033e-05,
+      "loss": 1.0918,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0440251572327045e-05,
+      "loss": 1.1013,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0566037735849058e-05,
+      "loss": 1.0938,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.069182389937107e-05,
+      "loss": 1.1086,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0817610062893083e-05,
+      "loss": 1.0505,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0943396226415095e-05,
+      "loss": 1.0698,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1069182389937107e-05,
+      "loss": 1.0562,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.119496855345912e-05,
+      "loss": 1.0913,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1320754716981132e-05,
+      "loss": 1.1279,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1446540880503145e-05,
+      "loss": 1.1128,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1572327044025157e-05,
+      "loss": 1.0383,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.169811320754717e-05,
+      "loss": 1.1021,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1823899371069182e-05,
+      "loss": 1.0669,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1949685534591196e-05,
+      "loss": 1.0872,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2075471698113209e-05,
+      "loss": 1.1526,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2201257861635221e-05,
+      "loss": 1.1023,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2327044025157234e-05,
+      "loss": 1.0356,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2452830188679246e-05,
+      "loss": 1.0525,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2578616352201259e-05,
+      "loss": 1.0359,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2704402515723271e-05,
+      "loss": 1.0857,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2830188679245283e-05,
+      "loss": 1.0818,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2955974842767296e-05,
+      "loss": 1.0623,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3081761006289308e-05,
+      "loss": 1.0725,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.320754716981132e-05,
+      "loss": 1.0432,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.1086,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3459119496855346e-05,
+      "loss": 1.1284,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3584905660377358e-05,
+      "loss": 1.0457,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.371069182389937e-05,
+      "loss": 1.0898,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3836477987421383e-05,
+      "loss": 1.0449,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3962264150943397e-05,
+      "loss": 1.0664,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.408805031446541e-05,
+      "loss": 1.0857,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4213836477987422e-05,
+      "loss": 1.0801,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4339622641509435e-05,
+      "loss": 1.0527,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4465408805031447e-05,
+      "loss": 1.0598,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.459119496855346e-05,
+      "loss": 1.0864,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4716981132075472e-05,
+      "loss": 1.0525,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4842767295597484e-05,
+      "loss": 1.0681,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4968553459119497e-05,
+      "loss": 1.0806,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5094339622641511e-05,
+      "loss": 1.053,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5220125786163525e-05,
+      "loss": 1.0923,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5345911949685536e-05,
+      "loss": 1.0532,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.547169811320755e-05,
+      "loss": 1.084,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.559748427672956e-05,
+      "loss": 1.0767,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5723270440251575e-05,
+      "loss": 1.0659,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5849056603773586e-05,
+      "loss": 1.0923,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.59748427672956e-05,
+      "loss": 1.0806,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.610062893081761e-05,
+      "loss": 1.0349,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6226415094339625e-05,
+      "loss": 1.0544,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.635220125786164e-05,
+      "loss": 1.0125,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.647798742138365e-05,
+      "loss": 1.0566,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6603773584905664e-05,
+      "loss": 1.0947,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6729559748427675e-05,
+      "loss": 1.0505,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.685534591194969e-05,
+      "loss": 1.0332,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.69811320754717e-05,
+      "loss": 1.0972,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7106918238993714e-05,
+      "loss": 1.0493,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7232704402515724e-05,
+      "loss": 1.0359,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.735849056603774e-05,
+      "loss": 1.0713,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.748427672955975e-05,
+      "loss": 1.0872,
+      "step": 139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7610062893081763e-05,
+      "loss": 1.1125,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7735849056603774e-05,
+      "loss": 1.0508,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7861635220125788e-05,
+      "loss": 1.0625,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.79874213836478e-05,
+      "loss": 1.0493,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8113207547169813e-05,
+      "loss": 1.0183,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8238993710691827e-05,
+      "loss": 1.0679,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8364779874213838e-05,
+      "loss": 1.0623,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8490566037735852e-05,
+      "loss": 1.0535,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8616352201257863e-05,
+      "loss": 1.0625,
+      "step": 148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8742138364779877e-05,
+      "loss": 1.051,
+      "step": 149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8867924528301888e-05,
+      "loss": 1.0854,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8993710691823902e-05,
+      "loss": 1.0813,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9119496855345913e-05,
+      "loss": 1.104,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9245283018867927e-05,
+      "loss": 1.0571,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9371069182389938e-05,
+      "loss": 1.0493,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9496855345911952e-05,
+      "loss": 1.0491,
+      "step": 155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9622641509433963e-05,
+      "loss": 1.0723,
+      "step": 156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9748427672955977e-05,
+      "loss": 1.072,
+      "step": 157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9874213836477987e-05,
+      "loss": 1.0886,
+      "step": 158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.0793,
+      "step": 159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999813141654e-05,
+      "loss": 1.0339,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999252566685e-05,
+      "loss": 1.0542,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999983182753032e-05,
+      "loss": 1.1143,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999970102678573e-05,
+      "loss": 1.0134,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999995328544836e-05,
+      "loss": 1.0725,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999932731068688e-05,
+      "loss": 1.0417,
+      "step": 165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999908439547223e-05,
+      "loss": 1.0474,
+      "step": 166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999880410893057e-05,
+      "loss": 1.0974,
+      "step": 167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999848645116656e-05,
+      "loss": 1.0442,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999813142229897e-05,
+      "loss": 1.0447,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999773902246042e-05,
+      "loss": 1.0261,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999973092517976e-05,
+      "loss": 1.0452,
+      "step": 171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999684211047113e-05,
+      "loss": 1.0859,
+      "step": 172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999633759865557e-05,
+      "loss": 1.0747,
+      "step": 173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999579571653944e-05,
+      "loss": 1.053,
+      "step": 174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999952164643253e-05,
+      "loss": 1.064,
+      "step": 175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999945998422296e-05,
+      "loss": 1.0359,
+      "step": 176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999939458504828e-05,
+      "loss": 1.0432,
+      "step": 177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999325448932926e-05,
+      "loss": 1.0386,
+      "step": 178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999252575902745e-05,
+      "loss": 1.0667,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999175965984957e-05,
+      "loss": 1.084,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999095619208202e-05,
+      "loss": 1.0554,
+      "step": 181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999011535602507e-05,
+      "loss": 1.0723,
+      "step": 182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999892371519929e-05,
+      "loss": 1.0476,
+      "step": 183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998832158031377e-05,
+      "loss": 1.0361,
+      "step": 184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998736864132978e-05,
+      "loss": 1.0603,
+      "step": 185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998637833539716e-05,
+      "loss": 1.0908,
+      "step": 186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999853506628859e-05,
+      "loss": 1.0769,
+      "step": 187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999842856241801e-05,
+      "loss": 1.0461,
+      "step": 188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998318321967774e-05,
+      "loss": 1.0449,
+      "step": 189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999820434497909e-05,
+      "loss": 1.0298,
+      "step": 190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998086631494547e-05,
+      "loss": 1.0205,
+      "step": 191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997965181558136e-05,
+      "loss": 1.0364,
+      "step": 192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997839995215246e-05,
+      "loss": 1.0713,
+      "step": 193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997711072512664e-05,
+      "loss": 1.0725,
+      "step": 194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997578413498567e-05,
+      "loss": 1.0337,
+      "step": 195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999744201822253e-05,
+      "loss": 1.0178,
+      "step": 196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997301886735535e-05,
+      "loss": 1.0486,
+      "step": 197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999715801908994e-05,
+      "loss": 1.0652,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997010415339518e-05,
+      "loss": 1.0449,
+      "step": 199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999685907553943e-05,
+      "loss": 1.1089,
+      "step": 200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996703999746234e-05,
+      "loss": 1.0413,
+      "step": 201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996545188017882e-05,
+      "loss": 1.0305,
+      "step": 202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999638264041373e-05,
+      "loss": 1.0474,
+      "step": 203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996216356994518e-05,
+      "loss": 1.0674,
+      "step": 204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999604633782239e-05,
+      "loss": 1.0601,
+      "step": 205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999587258296089e-05,
+      "loss": 1.0935,
+      "step": 206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995695092474953e-05,
+      "loss": 1.0417,
+      "step": 207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995513866430904e-05,
+      "loss": 1.0662,
+      "step": 208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995328904896473e-05,
+      "loss": 1.022,
+      "step": 209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995140207940785e-05,
+      "loss": 0.9993,
+      "step": 210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999494777563436e-05,
+      "loss": 1.0481,
+      "step": 211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994751608049106e-05,
+      "loss": 1.0232,
+      "step": 212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994551705258344e-05,
+      "loss": 1.0439,
+      "step": 213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994348067336775e-05,
+      "loss": 1.0625,
+      "step": 214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99941406943605e-05,
+      "loss": 1.1045,
+      "step": 215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999392958640703e-05,
+      "loss": 1.072,
+      "step": 216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999371474355524e-05,
+      "loss": 1.0154,
+      "step": 217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993496165885436e-05,
+      "loss": 1.0645,
+      "step": 218
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993273853479298e-05,
+      "loss": 1.0391,
+      "step": 219
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999304780641991e-05,
+      "loss": 0.9932,
+      "step": 220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999281802479175e-05,
+      "loss": 1.0688,
+      "step": 221
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992584508680685e-05,
+      "loss": 1.051,
+      "step": 222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992347258173993e-05,
+      "loss": 1.0549,
+      "step": 223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999210627336033e-05,
+      "loss": 1.0571,
+      "step": 224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999186155432976e-05,
+      "loss": 1.1123,
+      "step": 225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999161310117374e-05,
+      "loss": 1.0786,
+      "step": 226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991360913985122e-05,
+      "loss": 1.0496,
+      "step": 227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991104992858146e-05,
+      "loss": 1.0857,
+      "step": 228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999084533788846e-05,
+      "loss": 1.0945,
+      "step": 229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99905819491731e-05,
+      "loss": 1.0439,
+      "step": 230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99903148268105e-05,
+      "loss": 1.0605,
+      "step": 231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990043970900485e-05,
+      "loss": 1.0684,
+      "step": 232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989769381544278e-05,
+      "loss": 1.0652,
+      "step": 233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99894910588445e-05,
+      "loss": 1.0559,
+      "step": 234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989209002905167e-05,
+      "loss": 1.0493,
+      "step": 235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988923213831683e-05,
+      "loss": 1.0327,
+      "step": 236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988633691730856e-05,
+      "loss": 1.0205,
+      "step": 237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998834043671088e-05,
+      "loss": 1.0542,
+      "step": 238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988043448881356e-05,
+      "loss": 1.0762,
+      "step": 239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987742728353274e-05,
+      "loss": 1.0574,
+      "step": 240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987438275239008e-05,
+      "loss": 1.0632,
+      "step": 241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987130089652348e-05,
+      "loss": 1.0337,
+      "step": 242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998681817170846e-05,
+      "loss": 1.0256,
+      "step": 243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986502521523923e-05,
+      "loss": 1.053,
+      "step": 244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998618313921669e-05,
+      "loss": 1.0588,
+      "step": 245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985860024906128e-05,
+      "loss": 1.0383,
+      "step": 246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985533178712982e-05,
+      "loss": 1.063,
+      "step": 247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985202600759407e-05,
+      "loss": 1.0549,
+      "step": 248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984868291168943e-05,
+      "loss": 1.0276,
+      "step": 249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984530250066522e-05,
+      "loss": 1.041,
+      "step": 250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984188477578485e-05,
+      "loss": 1.0242,
+      "step": 251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998384297383255e-05,
+      "loss": 1.0454,
+      "step": 252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983493738957845e-05,
+      "loss": 1.0444,
+      "step": 253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998314077308488e-05,
+      "loss": 1.0356,
+      "step": 254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982784076345564e-05,
+      "loss": 1.0239,
+      "step": 255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99824236488732e-05,
+      "loss": 1.0417,
+      "step": 256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982059490802488e-05,
+      "loss": 1.0349,
+      "step": 257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998169160226952e-05,
+      "loss": 1.0671,
+      "step": 258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998131998341178e-05,
+      "loss": 1.0261,
+      "step": 259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980944634368152e-05,
+      "loss": 1.0251,
+      "step": 260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998056555527891e-05,
+      "loss": 1.0193,
+      "step": 261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980182746285717e-05,
+      "loss": 0.9951,
+      "step": 262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979796207531638e-05,
+      "loss": 1.0911,
+      "step": 263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997940593916113e-05,
+      "loss": 1.0745,
+      "step": 264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997901194132004e-05,
+      "loss": 1.0344,
+      "step": 265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978614214155615e-05,
+      "loss": 1.0562,
+      "step": 266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978212757816492e-05,
+      "loss": 0.9993,
+      "step": 267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99778075724527e-05,
+      "loss": 1.0273,
+      "step": 268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977398658215663e-05,
+      "loss": 1.02,
+      "step": 269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99769860152582e-05,
+      "loss": 1.0112,
+      "step": 270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976569643734526e-05,
+      "loss": 1.0796,
+      "step": 271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976149543800242e-05,
+      "loss": 1.021,
+      "step": 272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975725715612344e-05,
+      "loss": 1.0522,
+      "step": 273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975298159329228e-05,
+      "loss": 1.0398,
+      "step": 274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974866875110678e-05,
+      "loss": 1.0271,
+      "step": 275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997443186311787e-05,
+      "loss": 1.0623,
+      "step": 276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997399312351338e-05,
+      "loss": 1.0496,
+      "step": 277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997355065646117e-05,
+      "loss": 1.0322,
+      "step": 278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973104462126595e-05,
+      "loss": 1.0735,
+      "step": 279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972654540676408e-05,
+      "loss": 1.0369,
+      "step": 280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997220089227875e-05,
+      "loss": 1.0498,
+      "step": 281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997174351710316e-05,
+      "loss": 1.0107,
+      "step": 282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971282415320563e-05,
+      "loss": 1.0234,
+      "step": 283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970817587103284e-05,
+      "loss": 1.0142,
+      "step": 284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970349032625037e-05,
+      "loss": 1.0225,
+      "step": 285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969876752060924e-05,
+      "loss": 1.0342,
+      "step": 286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969400745587446e-05,
+      "loss": 1.0811,
+      "step": 287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99689210133825e-05,
+      "loss": 0.998,
+      "step": 288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968437555625364e-05,
+      "loss": 0.9907,
+      "step": 289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967950372496714e-05,
+      "loss": 1.0242,
+      "step": 290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996745946417862e-05,
+      "loss": 1.0657,
+      "step": 291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966964830854545e-05,
+      "loss": 1.0566,
+      "step": 292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996646647270934e-05,
+      "loss": 1.0698,
+      "step": 293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965964389929252e-05,
+      "loss": 1.0566,
+      "step": 294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996545858270191e-05,
+      "loss": 1.0549,
+      "step": 295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996494905121635e-05,
+      "loss": 1.0303,
+      "step": 296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964435795662988e-05,
+      "loss": 1.0618,
+      "step": 297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963918816233643e-05,
+      "loss": 1.0544,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996339811312151e-05,
+      "loss": 1.0791,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962873686521194e-05,
+      "loss": 1.0352,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996234553662867e-05,
+      "loss": 0.9849,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961813663641327e-05,
+      "loss": 1.0435,
+      "step": 302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996127806775793e-05,
+      "loss": 1.0459,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996073874917864e-05,
+      "loss": 1.0205,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960195708105014e-05,
+      "loss": 1.0513,
+      "step": 305
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995964894473999e-05,
+      "loss": 1.0076,
+      "step": 306
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959098459287906e-05,
+      "loss": 1.0178,
+      "step": 307
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958544251954483e-05,
+      "loss": 1.0559,
+      "step": 308
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9957986322946842e-05,
+      "loss": 1.0239,
+      "step": 309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995742467247349e-05,
+      "loss": 1.0508,
+      "step": 310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956859300744323e-05,
+      "loss": 1.0684,
+      "step": 311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995629020797063e-05,
+      "loss": 1.0613,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955717394365092e-05,
+      "loss": 1.0281,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955140860141776e-05,
+      "loss": 1.0945,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995456060551615e-05,
+      "loss": 1.0671,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995397663070506e-05,
+      "loss": 1.0464,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953388935926746e-05,
+      "loss": 1.0334,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952797521400836e-05,
+      "loss": 1.0225,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952202387348357e-05,
+      "loss": 1.0857,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951603533991724e-05,
+      "loss": 1.0796,
+      "step": 320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951000961554728e-05,
+      "loss": 1.0225,
+      "step": 321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995039467026257e-05,
+      "loss": 1.0149,
+      "step": 322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994978466034183e-05,
+      "loss": 0.9917,
+      "step": 323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949170932020472e-05,
+      "loss": 1.0205,
+      "step": 324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948553485527863e-05,
+      "loss": 1.071,
+      "step": 325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947932321094747e-05,
+      "loss": 1.0803,
+      "step": 326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994730743895327e-05,
+      "loss": 1.0276,
+      "step": 327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946678839336958e-05,
+      "loss": 1.0635,
+      "step": 328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946046522480728e-05,
+      "loss": 1.0273,
+      "step": 329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994541048862089e-05,
+      "loss": 1.062,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944770737995142e-05,
+      "loss": 0.9849,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994412727084256e-05,
+      "loss": 1.0349,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943480087403634e-05,
+      "loss": 1.071,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994282918792021e-05,
+      "loss": 1.0457,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942174572635552e-05,
+      "loss": 1.0618,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941516241794295e-05,
+      "loss": 1.0706,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940854195642478e-05,
+      "loss": 1.0356,
+      "step": 337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99401884344275e-05,
+      "loss": 1.0105,
+      "step": 338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939518958398187e-05,
+      "loss": 1.0425,
+      "step": 339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993884576780472e-05,
+      "loss": 1.0588,
+      "step": 340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938168862898685e-05,
+      "loss": 1.0537,
+      "step": 341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937488243933055e-05,
+      "loss": 1.0391,
+      "step": 342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993680391116219e-05,
+      "loss": 0.9736,
+      "step": 343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993611586484183e-05,
+      "loss": 1.0652,
+      "step": 344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935424105229117e-05,
+      "loss": 1.0557,
+      "step": 345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934728632582566e-05,
+      "loss": 1.0334,
+      "step": 346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993402944716209e-05,
+      "loss": 1.0688,
+      "step": 347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993332654922899e-05,
+      "loss": 1.0242,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993261993904594e-05,
+      "loss": 1.0413,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993190961687703e-05,
+      "loss": 1.0034,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931195582987698e-05,
+      "loss": 1.0554,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930477837644806e-05,
+      "loss": 1.0483,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929756381116582e-05,
+      "loss": 1.0461,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929031213672646e-05,
+      "loss": 1.0266,
+      "step": 354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928302335584004e-05,
+      "loss": 1.0464,
+      "step": 355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927569747123054e-05,
+      "loss": 1.0105,
+      "step": 356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926833448563574e-05,
+      "loss": 1.0422,
+      "step": 357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992609344018073e-05,
+      "loss": 1.0239,
+      "step": 358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925349722251077e-05,
+      "loss": 1.0872,
+      "step": 359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924602295052554e-05,
+      "loss": 1.0581,
+      "step": 360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923851158864485e-05,
+      "loss": 1.0232,
+      "step": 361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992309631396759e-05,
+      "loss": 1.011,
+      "step": 362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992233776064396e-05,
+      "loss": 1.0569,
+      "step": 363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992157549917708e-05,
+      "loss": 1.0115,
+      "step": 364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992080952985182e-05,
+      "loss": 0.9944,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920039852954435e-05,
+      "loss": 1.0479,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991926646877257e-05,
+      "loss": 1.0447,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991848937759525e-05,
+      "loss": 1.0071,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917708579712882e-05,
+      "loss": 1.0771,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991692407541727e-05,
+      "loss": 1.0613,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916135865001592e-05,
+      "loss": 1.0334,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915343948760418e-05,
+      "loss": 1.0244,
+      "step": 372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99145483269897e-05,
+      "loss": 1.0576,
+      "step": 373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913748999986772e-05,
+      "loss": 1.0291,
+      "step": 374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912945968050357e-05,
+      "loss": 1.0776,
+      "step": 375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912139231480564e-05,
+      "loss": 1.053,
+      "step": 376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911328790578883e-05,
+      "loss": 0.9849,
+      "step": 377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910514645648188e-05,
+      "loss": 1.0007,
+      "step": 378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909696796992737e-05,
+      "loss": 1.051,
+      "step": 379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990887524491818e-05,
+      "loss": 1.0469,
+      "step": 380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908049989731538e-05,
+      "loss": 1.0371,
+      "step": 381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907221031741227e-05,
+      "loss": 1.0454,
+      "step": 382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906388371257038e-05,
+      "loss": 1.002,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905552008590154e-05,
+      "loss": 1.0315,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904711944053137e-05,
+      "loss": 1.0403,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990386817795993e-05,
+      "loss": 1.0601,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903020710625874e-05,
+      "loss": 0.9888,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902169542367664e-05,
+      "loss": 1.0171,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990131467350341e-05,
+      "loss": 1.0081,
+      "step": 389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900456104352582e-05,
+      "loss": 1.0864,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899593835236048e-05,
+      "loss": 1.0007,
+      "step": 391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898727866476047e-05,
+      "loss": 0.999,
+      "step": 392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897858198396207e-05,
+      "loss": 1.0315,
+      "step": 393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896984831321542e-05,
+      "loss": 1.0134,
+      "step": 394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896107765578437e-05,
+      "loss": 1.0361,
+      "step": 395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895227001494674e-05,
+      "loss": 1.0408,
+      "step": 396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894342539399402e-05,
+      "loss": 1.0547,
+      "step": 397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893454379623165e-05,
+      "loss": 1.0254,
+      "step": 398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892562522497877e-05,
+      "loss": 1.0146,
+      "step": 399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891666968356846e-05,
+      "loss": 1.0234,
+      "step": 400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890767717534755e-05,
+      "loss": 1.0256,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889864770367662e-05,
+      "loss": 1.0183,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888958127193023e-05,
+      "loss": 1.061,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888047788349656e-05,
+      "loss": 1.0859,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887133754177775e-05,
+      "loss": 1.0417,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988621602501897e-05,
+      "loss": 1.0144,
+      "step": 406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885294601216215e-05,
+      "loss": 0.9924,
+      "step": 407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884369483113858e-05,
+      "loss": 1.0171,
+      "step": 408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988344067105763e-05,
+      "loss": 1.0417,
+      "step": 409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988250816539464e-05,
+      "loss": 1.0725,
+      "step": 410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881571966473392e-05,
+      "loss": 1.0452,
+      "step": 411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988063207464375e-05,
+      "loss": 0.9858,
+      "step": 412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879688490256967e-05,
+      "loss": 1.0454,
+      "step": 413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878741213665683e-05,
+      "loss": 1.0378,
+      "step": 414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9877790245223906e-05,
+      "loss": 1.0398,
+      "step": 415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987683558528703e-05,
+      "loss": 1.0459,
+      "step": 416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875877234211828e-05,
+      "loss": 1.02,
+      "step": 417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987491519235645e-05,
+      "loss": 0.9966,
+      "step": 418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873949460080423e-05,
+      "loss": 1.0344,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872980037744672e-05,
+      "loss": 1.0269,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872006925711468e-05,
+      "loss": 1.0151,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871030124344496e-05,
+      "loss": 1.0332,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987004963400879e-05,
+      "loss": 1.0396,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869065455070782e-05,
+      "loss": 1.0417,
+      "step": 424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868077587898275e-05,
+      "loss": 1.0168,
+      "step": 425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986708603286045e-05,
+      "loss": 1.0344,
+      "step": 426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866090790327868e-05,
+      "loss": 1.0107,
+      "step": 427
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986509186067247e-05,
+      "loss": 0.9893,
+      "step": 428
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986408924426757e-05,
+      "loss": 1.0024,
+      "step": 429
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863082941487865e-05,
+      "loss": 0.9951,
+      "step": 430
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986207295270943e-05,
+      "loss": 1.0286,
+      "step": 431
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861059278309703e-05,
+      "loss": 0.9861,
+      "step": 432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.986004191866752e-05,
+      "loss": 1.042,
+      "step": 433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859020874163086e-05,
+      "loss": 1.0093,
+      "step": 434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985799614517798e-05,
+      "loss": 1.0312,
+      "step": 435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985696773209516e-05,
+      "loss": 1.072,
+      "step": 436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855935635298963e-05,
+      "loss": 1.009,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854899855175098e-05,
+      "loss": 0.9915,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853860392110654e-05,
+      "loss": 1.0208,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98528172464941e-05,
+      "loss": 0.9756,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985177041871527e-05,
+      "loss": 1.0928,
+      "step": 441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850719909165384e-05,
+      "loss": 1.0891,
+      "step": 442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849665718237036e-05,
+      "loss": 0.9712,
+      "step": 443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848607846324195e-05,
+      "loss": 0.9795,
+      "step": 444
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847546293822206e-05,
+      "loss": 1.0723,
+      "step": 445
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846481061127788e-05,
+      "loss": 0.9983,
+      "step": 446
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845412148639033e-05,
+      "loss": 1.0859,
+      "step": 447
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844339556755413e-05,
+      "loss": 0.9885,
+      "step": 448
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843263285877776e-05,
+      "loss": 1.0654,
+      "step": 449
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842183336408342e-05,
+      "loss": 1.0195,
+      "step": 450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841099708750705e-05,
+      "loss": 1.0356,
+      "step": 451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840012403309834e-05,
+      "loss": 1.0286,
+      "step": 452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838921420492073e-05,
+      "loss": 1.0632,
+      "step": 453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837826760705144e-05,
+      "loss": 1.043,
+      "step": 454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836728424358136e-05,
+      "loss": 1.063,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835626411861515e-05,
+      "loss": 1.0537,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834520723627127e-05,
+      "loss": 1.0476,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983341136006818e-05,
+      "loss": 1.0239,
+      "step": 458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832298321599265e-05,
+      "loss": 0.9814,
+      "step": 459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831181608636343e-05,
+      "loss": 0.9941,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830061221596746e-05,
+      "loss": 1.0559,
+      "step": 461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982893716089918e-05,
+      "loss": 1.0332,
+      "step": 462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827809426963733e-05,
+      "loss": 1.0349,
+      "step": 463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982667802021185e-05,
+      "loss": 1.0378,
+      "step": 464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982554294106636e-05,
+      "loss": 1.0649,
+      "step": 465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824404189951462e-05,
+      "loss": 0.9861,
+      "step": 466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823261767292724e-05,
+      "loss": 1.04,
+      "step": 467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982211567351709e-05,
+      "loss": 1.0464,
+      "step": 468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820965909052876e-05,
+      "loss": 1.0144,
+      "step": 469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819812474329762e-05,
+      "loss": 0.9915,
+      "step": 470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981865536977881e-05,
+      "loss": 1.0295,
+      "step": 471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981749459583245e-05,
+      "loss": 1.0068,
+      "step": 472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816330152924484e-05,
+      "loss": 1.002,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981516204149008e-05,
+      "loss": 0.9995,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813990261965776e-05,
+      "loss": 1.0208,
+      "step": 475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98128148147895e-05,
+      "loss": 1.0317,
+      "step": 476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811635700400523e-05,
+      "loss": 1.0168,
+      "step": 477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810452919239506e-05,
+      "loss": 0.9783,
+      "step": 478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809266471748475e-05,
+      "loss": 1.0212,
+      "step": 479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980807635837082e-05,
+      "loss": 1.0493,
+      "step": 480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980688257955131e-05,
+      "loss": 1.01,
+      "step": 481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805685135736076e-05,
+      "loss": 1.02,
+      "step": 482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980448402737263e-05,
+      "loss": 1.0293,
+      "step": 483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803279254909842e-05,
+      "loss": 0.979,
+      "step": 484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802070818797957e-05,
+      "loss": 1.042,
+      "step": 485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980085871948858e-05,
+      "loss": 1.0586,
+      "step": 486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799642957434707e-05,
+      "loss": 1.0405,
+      "step": 487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798423533090675e-05,
+      "loss": 1.0188,
+      "step": 488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797200446912216e-05,
+      "loss": 0.9858,
+      "step": 489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795973699356403e-05,
+      "loss": 1.022,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794743290881705e-05,
+      "loss": 1.0576,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979350922194794e-05,
+      "loss": 1.0171,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97922714930163e-05,
+      "loss": 1.0933,
+      "step": 493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791030104549347e-05,
+      "loss": 1.0254,
+      "step": 494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789785057011006e-05,
+      "loss": 0.9912,
+      "step": 495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788536350866577e-05,
+      "loss": 1.0696,
+      "step": 496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787283986582716e-05,
+      "loss": 1.0054,
+      "step": 497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786027964627457e-05,
+      "loss": 1.0049,
+      "step": 498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784768285470195e-05,
+      "loss": 1.0168,
+      "step": 499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783504949581694e-05,
+      "loss": 1.0203,
+      "step": 500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978223795743408e-05,
+      "loss": 0.9863,
+      "step": 501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780967309500854e-05,
+      "loss": 1.0327,
+      "step": 502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9779693006256874e-05,
+      "loss": 1.0981,
+      "step": 503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778415048178375e-05,
+      "loss": 1.042,
+      "step": 504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777133435742945e-05,
+      "loss": 1.0312,
+      "step": 505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775848169429547e-05,
+      "loss": 1.0715,
+      "step": 506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97745592497185e-05,
+      "loss": 1.0286,
+      "step": 507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773266677091503e-05,
+      "loss": 1.0425,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771970452031612e-05,
+      "loss": 1.0208,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770670575023243e-05,
+      "loss": 0.9905,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769367046552183e-05,
+      "loss": 0.9988,
+      "step": 511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768059867105584e-05,
+      "loss": 1.0459,
+      "step": 512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976674903717196e-05,
+      "loss": 1.0461,
+      "step": 513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976543455724119e-05,
+      "loss": 1.0332,
+      "step": 514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764116427804513e-05,
+      "loss": 1.0081,
+      "step": 515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762794649354546e-05,
+      "loss": 1.0413,
+      "step": 516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976146922238525e-05,
+      "loss": 1.0132,
+      "step": 517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760140147391963e-05,
+      "loss": 0.9978,
+      "step": 518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758807424871385e-05,
+      "loss": 0.9917,
+      "step": 519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975747105532157e-05,
+      "loss": 1.0322,
+      "step": 520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975613103924195e-05,
+      "loss": 1.063,
+      "step": 521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754787377133306e-05,
+      "loss": 1.0237,
+      "step": 522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753440069497785e-05,
+      "loss": 0.9993,
+      "step": 523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752089116838905e-05,
+      "loss": 1.0098,
+      "step": 524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750734519661534e-05,
+      "loss": 1.0449,
+      "step": 525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974937627847191e-05,
+      "loss": 0.9753,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974801439377763e-05,
+      "loss": 1.0415,
+      "step": 527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974664886608765e-05,
+      "loss": 1.0342,
+      "step": 528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97452796959123e-05,
+      "loss": 1.0481,
+      "step": 529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743906883763246e-05,
+      "loss": 1.0303,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742530430153548e-05,
+      "loss": 1.041,
+      "step": 531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741150335597598e-05,
+      "loss": 1.0273,
+      "step": 532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973976660061116e-05,
+      "loss": 0.9993,
+      "step": 533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973837922571137e-05,
+      "loss": 0.9976,
+      "step": 534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736988211416704e-05,
+      "loss": 0.9998,
+      "step": 535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735593558247007e-05,
+      "loss": 1.0039,
+      "step": 536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973419526672349e-05,
+      "loss": 1.0225,
+      "step": 537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732793337368713e-05,
+      "loss": 1.0144,
+      "step": 538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97313877707066e-05,
+      "loss": 1.0422,
+      "step": 539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972997856726244e-05,
+      "loss": 1.009,
+      "step": 540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972856572756287e-05,
+      "loss": 0.9817,
+      "step": 541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727149252135894e-05,
+      "loss": 1.0464,
+      "step": 542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725729141510873e-05,
+      "loss": 0.9868,
+      "step": 543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724305396218523e-05,
+      "loss": 1.0193,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722878016790928e-05,
+      "loss": 1.0142,
+      "step": 545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721447003761517e-05,
+      "loss": 0.9998,
+      "step": 546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720012357665087e-05,
+      "loss": 1.0007,
+      "step": 547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971857407903779e-05,
+      "loss": 0.9866,
+      "step": 548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717132168417132e-05,
+      "loss": 1.041,
+      "step": 549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971568662634198e-05,
+      "loss": 1.0474,
+      "step": 550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714237453352555e-05,
+      "loss": 1.0059,
+      "step": 551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712784649990442e-05,
+      "loss": 1.0093,
+      "step": 552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971132821679857e-05,
+      "loss": 1.062,
+      "step": 553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709868154321246e-05,
+      "loss": 0.9792,
+      "step": 554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708404463104104e-05,
+      "loss": 1.0342,
+      "step": 555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970693714369416e-05,
+      "loss": 1.0193,
+      "step": 556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705466196639773e-05,
+      "loss": 1.0183,
+      "step": 557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970399162249066e-05,
+      "loss": 1.021,
+      "step": 558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702513421797897e-05,
+      "loss": 1.0469,
+      "step": 559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970103159511391e-05,
+      "loss": 1.0464,
+      "step": 560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699546142992476e-05,
+      "loss": 1.0149,
+      "step": 561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969805706598874e-05,
+      "loss": 1.0405,
+      "step": 562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696564364659202e-05,
+      "loss": 1.032,
+      "step": 563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695068039561697e-05,
+      "loss": 1.0254,
+      "step": 564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969356809125543e-05,
+      "loss": 1.0085,
+      "step": 565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692064520300958e-05,
+      "loss": 1.0308,
+      "step": 566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969055732726019e-05,
+      "loss": 0.989,
+      "step": 567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968904651269639e-05,
+      "loss": 1.0115,
+      "step": 568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687532077174176e-05,
+      "loss": 1.0059,
+      "step": 569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968601402125952e-05,
+      "loss": 1.0308,
+      "step": 570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684492345519734e-05,
+      "loss": 1.0154,
+      "step": 571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682967050523504e-05,
+      "loss": 0.9951,
+      "step": 572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681438136840854e-05,
+      "loss": 1.0422,
+      "step": 573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9679905605043167e-05,
+      "loss": 1.0408,
+      "step": 574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967836945570317e-05,
+      "loss": 1.0479,
+      "step": 575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676829689394956e-05,
+      "loss": 1.0012,
+      "step": 576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675286306693957e-05,
+      "loss": 1.0496,
+      "step": 577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673739308176957e-05,
+      "loss": 1.0269,
+      "step": 578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672188694422105e-05,
+      "loss": 1.0264,
+      "step": 579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967063446600888e-05,
+      "loss": 1.0427,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669076623518132e-05,
+      "loss": 1.0537,
+      "step": 581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966751516753205e-05,
+      "loss": 0.9963,
+      "step": 582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665950098634172e-05,
+      "loss": 1.0764,
+      "step": 583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664381417409396e-05,
+      "loss": 1.0334,
+      "step": 584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662809124443965e-05,
+      "loss": 1.0376,
+      "step": 585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966123322032546e-05,
+      "loss": 0.9983,
+      "step": 586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965965370564284e-05,
+      "loss": 1.0464,
+      "step": 587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658070580986382e-05,
+      "loss": 1.0244,
+      "step": 588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656483846947733e-05,
+      "loss": 1.0037,
+      "step": 589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965489350411988e-05,
+      "loss": 1.031,
+      "step": 590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965329955309716e-05,
+      "loss": 1.0181,
+      "step": 591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965170199447526e-05,
+      "loss": 1.0444,
+      "step": 592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650100828851216e-05,
+      "loss": 0.9883,
+      "step": 593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648496056823404e-05,
+      "loss": 1.0161,
+      "step": 594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646887678991562e-05,
+      "loss": 0.9736,
+      "step": 595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645275695956765e-05,
+      "loss": 0.9995,
+      "step": 596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643660108321436e-05,
+      "loss": 0.9783,
+      "step": 597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642040916689348e-05,
+      "loss": 1.0315,
+      "step": 598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640418121665622e-05,
+      "loss": 1.0234,
+      "step": 599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963879172385672e-05,
+      "loss": 1.0317,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637161723870456e-05,
+      "loss": 1.0073,
+      "step": 601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635528122315988e-05,
+      "loss": 0.9993,
+      "step": 602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963389091980382e-05,
+      "loss": 1.0425,
+      "step": 603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632250116945802e-05,
+      "loss": 1.0273,
+      "step": 604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963060571435513e-05,
+      "loss": 1.0208,
+      "step": 605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628957712646346e-05,
+      "loss": 0.9778,
+      "step": 606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962730611243533e-05,
+      "loss": 1.0125,
+      "step": 607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962565091433932e-05,
+      "loss": 0.988,
+      "step": 608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962399211897689e-05,
+      "loss": 1.0264,
+      "step": 609
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962232972696795e-05,
+      "loss": 1.0261,
+      "step": 610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620663738933775e-05,
+      "loss": 1.0068,
+      "step": 611
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961899415549697e-05,
+      "loss": 1.0005,
+      "step": 612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617320977281482e-05,
+      "loss": 1.0176,
+      "step": 613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615644204912607e-05,
+      "loss": 1.0432,
+      "step": 614
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613963839016988e-05,
+      "loss": 0.9946,
+      "step": 615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612279880222596e-05,
+      "loss": 1.0349,
+      "step": 616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610592329158765e-05,
+      "loss": 1.0388,
+      "step": 617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608901186456157e-05,
+      "loss": 1.0146,
+      "step": 618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607206452746782e-05,
+      "loss": 1.0051,
+      "step": 619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605508128663982e-05,
+      "loss": 1.0469,
+      "step": 620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603806214842457e-05,
+      "loss": 1.0396,
+      "step": 621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602100711918236e-05,
+      "loss": 1.0105,
+      "step": 622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600391620528702e-05,
+      "loss": 1.0188,
+      "step": 623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598678941312562e-05,
+      "loss": 1.0305,
+      "step": 624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596962674909877e-05,
+      "loss": 1.0251,
+      "step": 625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595242821962042e-05,
+      "loss": 0.99,
+      "step": 626
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.95935193831118e-05,
+      "loss": 0.9768,
+      "step": 627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959179235900322e-05,
+      "loss": 0.9785,
+      "step": 628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959006175028173e-05,
+      "loss": 0.9861,
+      "step": 629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958832755759408e-05,
+      "loss": 1.0505,
+      "step": 630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958658978158837e-05,
+      "loss": 0.9944,
+      "step": 631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584848422914034e-05,
+      "loss": 1.0613,
+      "step": 632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583103482221848e-05,
+      "loss": 0.998,
+      "step": 633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581354960163924e-05,
+      "loss": 1.011,
+      "step": 634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957960285739372e-05,
+      "loss": 0.9458,
+      "step": 635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577847174566016e-05,
+      "loss": 1.0208,
+      "step": 636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576087912336946e-05,
+      "loss": 0.9927,
+      "step": 637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574325071363978e-05,
+      "loss": 1.0396,
+      "step": 638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572558652305912e-05,
+      "loss": 1.0361,
+      "step": 639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570788655822886e-05,
+      "loss": 1.0393,
+      "step": 640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569015082576378e-05,
+      "loss": 1.0481,
+      "step": 641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567237933229208e-05,
+      "loss": 1.0195,
+      "step": 642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565457208445518e-05,
+      "loss": 0.9893,
+      "step": 643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9563672908890798e-05,
+      "loss": 0.9802,
+      "step": 644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561885035231873e-05,
+      "loss": 1.0388,
+      "step": 645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95600935881369e-05,
+      "loss": 1.0369,
+      "step": 646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955829856827537e-05,
+      "loss": 0.9668,
+      "step": 647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556499976318116e-05,
+      "loss": 1.0276,
+      "step": 648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95546978129373e-05,
+      "loss": 0.9304,
+      "step": 649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552892078806413e-05,
+      "loss": 1.0583,
+      "step": 650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551082774600304e-05,
+      "loss": 1.0156,
+      "step": 651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549269900995124e-05,
+      "loss": 1.0173,
+      "step": 652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547453458668385e-05,
+      "loss": 1.0491,
+      "step": 653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545633448298918e-05,
+      "loss": 0.9958,
+      "step": 654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954380987056689e-05,
+      "loss": 1.0334,
+      "step": 655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541982726153805e-05,
+      "loss": 0.9854,
+      "step": 656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95401520157425e-05,
+      "loss": 1.0261,
+      "step": 657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538317740017133e-05,
+      "loss": 1.0068,
+      "step": 658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536479899663208e-05,
+      "loss": 0.9854,
+      "step": 659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534638495367563e-05,
+      "loss": 1.0222,
+      "step": 660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532793527818354e-05,
+      "loss": 1.0474,
+      "step": 661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9530944997705075e-05,
+      "loss": 1.0132,
+      "step": 662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529092905718555e-05,
+      "loss": 1.0244,
+      "step": 663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527237252550953e-05,
+      "loss": 1.0479,
+      "step": 664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952537803889576e-05,
+      "loss": 1.0164,
+      "step": 665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952351526544779e-05,
+      "loss": 1.0261,
+      "step": 666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521648932903197e-05,
+      "loss": 1.0256,
+      "step": 667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519779041959456e-05,
+      "loss": 1.0654,
+      "step": 668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951790559331538e-05,
+      "loss": 1.0273,
+      "step": 669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516028587671104e-05,
+      "loss": 1.0281,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514148025728102e-05,
+      "loss": 0.9915,
+      "step": 671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512263908189165e-05,
+      "loss": 0.9993,
+      "step": 672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510376235758427e-05,
+      "loss": 1.0281,
+      "step": 673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9508485009141336e-05,
+      "loss": 1.0076,
+      "step": 674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506590229044678e-05,
+      "loss": 1.0203,
+      "step": 675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504691896176564e-05,
+      "loss": 1.0312,
+      "step": 676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502790011246427e-05,
+      "loss": 1.0195,
+      "step": 677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500884574965044e-05,
+      "loss": 1.0222,
+      "step": 678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.94989755880445e-05,
+      "loss": 1.0444,
+      "step": 679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9497063051198218e-05,
+      "loss": 1.0149,
+      "step": 680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495146965140943e-05,
+      "loss": 0.9971,
+      "step": 681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493227330588752e-05,
+      "loss": 1.0098,
+      "step": 682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949130414825904e-05,
+      "loss": 0.989,
+      "step": 683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489377418870536e-05,
+      "loss": 1.0173,
+      "step": 684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948744714314329e-05,
+      "loss": 0.9998,
+      "step": 685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948551332179868e-05,
+      "loss": 0.9976,
+      "step": 686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483575955559405e-05,
+      "loss": 1.0154,
+      "step": 687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481635045149488e-05,
+      "loss": 1.0205,
+      "step": 688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9479690591294285e-05,
+      "loss": 1.0088,
+      "step": 689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477742594720467e-05,
+      "loss": 0.9937,
+      "step": 690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9475791056156038e-05,
+      "loss": 0.939,
+      "step": 691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473835976330316e-05,
+      "loss": 0.9929,
+      "step": 692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471877355973947e-05,
+      "loss": 1.0059,
+      "step": 693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469915195818902e-05,
+      "loss": 1.0427,
+      "step": 694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946794949659847e-05,
+      "loss": 0.9561,
+      "step": 695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946598025904727e-05,
+      "loss": 0.9971,
+      "step": 696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946400748390124e-05,
+      "loss": 1.0403,
+      "step": 697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9462031171897635e-05,
+      "loss": 1.0076,
+      "step": 698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460051323775033e-05,
+      "loss": 1.002,
+      "step": 699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945806794027334e-05,
+      "loss": 0.9822,
+      "step": 700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456081022133784e-05,
+      "loss": 0.9941,
+      "step": 701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94540905700989e-05,
+      "loss": 1.0144,
+      "step": 702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945209658491256e-05,
+      "loss": 1.0168,
+      "step": 703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945009906731995e-05,
+      "loss": 1.0027,
+      "step": 704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944809801806757e-05,
+      "loss": 0.9802,
+      "step": 705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944609343790325e-05,
+      "loss": 0.9778,
+      "step": 706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444085327576133e-05,
+      "loss": 1.0667,
+      "step": 707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9442073687836687e-05,
+      "loss": 1.0564,
+      "step": 708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944005851943669e-05,
+      "loss": 1.0386,
+      "step": 709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943803982312925e-05,
+      "loss": 1.0178,
+      "step": 710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943601759966878e-05,
+      "loss": 0.9783,
+      "step": 711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433991849811025e-05,
+      "loss": 0.9976,
+      "step": 712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9431962574313037e-05,
+      "loss": 1.0276,
+      "step": 713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9429929773933195e-05,
+      "loss": 1.0127,
+      "step": 714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9427893449431184e-05,
+      "loss": 0.967,
+      "step": 715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425853601568025e-05,
+      "loss": 0.9641,
+      "step": 716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423810231106027e-05,
+      "loss": 1.0039,
+      "step": 717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421763338808842e-05,
+      "loss": 1.0164,
+      "step": 718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419712925441424e-05,
+      "loss": 0.9854,
+      "step": 719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417658991770048e-05,
+      "loss": 1.0154,
+      "step": 720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415601538562305e-05,
+      "loss": 1.0312,
+      "step": 721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413540566587093e-05,
+      "loss": 1.0054,
+      "step": 722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941147607661464e-05,
+      "loss": 0.9902,
+      "step": 723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409408069416477e-05,
+      "loss": 1.0181,
+      "step": 724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9407336545765454e-05,
+      "loss": 1.0359,
+      "step": 725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940526150643573e-05,
+      "loss": 1.0278,
+      "step": 726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403182952202782e-05,
+      "loss": 1.0071,
+      "step": 727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401100883843406e-05,
+      "loss": 1.0115,
+      "step": 728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.93990153021357e-05,
+      "loss": 1.022,
+      "step": 729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.9863,
+      "step": 730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394833601794286e-05,
+      "loss": 1.0115,
+      "step": 731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939273748472335e-05,
+      "loss": 0.9814,
+      "step": 732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9390637857429628e-05,
+      "loss": 1.0042,
+      "step": 733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388534720697785e-05,
+      "loss": 1.0132,
+      "step": 734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93864280753138e-05,
+      "loss": 0.9937,
+      "step": 735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384317922064962e-05,
+      "loss": 1.009,
+      "step": 736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938220426173987e-05,
+      "loss": 1.0244,
+      "step": 737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938008709512843e-05,
+      "loss": 1.0183,
+      "step": 738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377966423021868e-05,
+      "loss": 1.0195,
+      "step": 739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937584224621271e-05,
+      "loss": 0.9946,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373714565494804e-05,
+      "loss": 1.0398,
+      "step": 741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937158338166329e-05,
+      "loss": 1.0586,
+      "step": 742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369448695514636e-05,
+      "loss": 1.0103,
+      "step": 743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93673105078466e-05,
+      "loss": 1.0073,
+      "step": 744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936516881945827e-05,
+      "loss": 1.01,
+      "step": 745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936302363115002e-05,
+      "loss": 1.0415,
+      "step": 746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936087494372355e-05,
+      "loss": 0.9951,
+      "step": 747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358722757981854e-05,
+      "loss": 0.989,
+      "step": 748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356567074729248e-05,
+      "loss": 0.9883,
+      "step": 749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935440789477134e-05,
+      "loss": 1.0276,
+      "step": 750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352245218915053e-05,
+      "loss": 0.9873,
+      "step": 751
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350079047968616e-05,
+      "loss": 0.9614,
+      "step": 752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9347909382741564e-05,
+      "loss": 0.991,
+      "step": 753
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345736224044736e-05,
+      "loss": 1.0066,
+      "step": 754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934355957269028e-05,
+      "loss": 1.0049,
+      "step": 755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9341379429491644e-05,
+      "loss": 1.0007,
+      "step": 756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933919579526358e-05,
+      "loss": 0.98,
+      "step": 757
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933700867082216e-05,
+      "loss": 1.0234,
+      "step": 758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933481805698474e-05,
+      "loss": 1.0657,
+      "step": 759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933262395456999e-05,
+      "loss": 1.0444,
+      "step": 760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330426364397885e-05,
+      "loss": 1.0156,
+      "step": 761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93282252872897e-05,
+      "loss": 0.9861,
+      "step": 762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326020724068015e-05,
+      "loss": 0.9851,
+      "step": 763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323812675556706e-05,
+      "loss": 1.0427,
+      "step": 764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932160114258097e-05,
+      "loss": 0.988,
+      "step": 765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319386125967282e-05,
+      "loss": 1.0229,
+      "step": 766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931716762654344e-05,
+      "loss": 0.9758,
+      "step": 767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9314945645138524e-05,
+      "loss": 1.0334,
+      "step": 768
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312720182582936e-05,
+      "loss": 0.9783,
+      "step": 769
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310491239708362e-05,
+      "loss": 1.0642,
+      "step": 770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308258817347795e-05,
+      "loss": 1.0073,
+      "step": 771
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306022916335534e-05,
+      "loss": 1.0237,
+      "step": 772
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303783537507166e-05,
+      "loss": 1.0408,
+      "step": 773
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930154068169959e-05,
+      "loss": 1.0103,
+      "step": 774
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299294349750994e-05,
+      "loss": 0.9944,
+      "step": 775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929704454250087e-05,
+      "loss": 1.01,
+      "step": 776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9294791260790012e-05,
+      "loss": 1.0264,
+      "step": 777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929253450546051e-05,
+      "loss": 1.0161,
+      "step": 778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290274277355744e-05,
+      "loss": 1.0125,
+      "step": 779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9288010577320408e-05,
+      "loss": 0.9604,
+      "step": 780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285743406200472e-05,
+      "loss": 1.01,
+      "step": 781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928347276484323e-05,
+      "loss": 0.9751,
+      "step": 782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928119865409725e-05,
+      "loss": 1.0178,
+      "step": 783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278921074812403e-05,
+      "loss": 1.0144,
+      "step": 784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927664002783987e-05,
+      "loss": 1.0295,
+      "step": 785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9274355514032105e-05,
+      "loss": 0.9963,
+      "step": 786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927206753424287e-05,
+      "loss": 1.0044,
+      "step": 787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269776089327228e-05,
+      "loss": 1.0098,
+      "step": 788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267481180141525e-05,
+      "loss": 1.0215,
+      "step": 789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265182807543408e-05,
+      "loss": 0.9988,
+      "step": 790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262880972391818e-05,
+      "loss": 1.0601,
+      "step": 791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260575675546985e-05,
+      "loss": 0.9878,
+      "step": 792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258266917870447e-05,
+      "loss": 1.0369,
+      "step": 793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255954700225015e-05,
+      "loss": 1.0557,
+      "step": 794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9253639023474805e-05,
+      "loss": 0.9646,
+      "step": 795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925131988848523e-05,
+      "loss": 0.9612,
+      "step": 796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9248997296122983e-05,
+      "loss": 1.0151,
+      "step": 797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924667124725606e-05,
+      "loss": 1.0388,
+      "step": 798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924434174275374e-05,
+      "loss": 0.9402,
+      "step": 799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9242008783486603e-05,
+      "loss": 0.9634,
+      "step": 800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923967237032651e-05,
+      "loss": 0.9756,
+      "step": 801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923733250414662e-05,
+      "loss": 0.9712,
+      "step": 802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923498918582138e-05,
+      "loss": 0.9797,
+      "step": 803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232642416226526e-05,
+      "loss": 1.0034,
+      "step": 804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230292196239088e-05,
+      "loss": 1.0073,
+      "step": 805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9227938526737376e-05,
+      "loss": 1.0061,
+      "step": 806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9225581408601002e-05,
+      "loss": 1.0212,
+      "step": 807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9223220842710857e-05,
+      "loss": 0.9836,
+      "step": 808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9220856829949127e-05,
+      "loss": 1.02,
+      "step": 809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921848937119928e-05,
+      "loss": 0.9683,
+      "step": 810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9216118467346075e-05,
+      "loss": 1.0098,
+      "step": 811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921374411927556e-05,
+      "loss": 1.0674,
+      "step": 812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211366327875067e-05,
+      "loss": 0.9966,
+      "step": 813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920898509403322e-05,
+      "loss": 0.9902,
+      "step": 814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920660041863992e-05,
+      "loss": 1.0066,
+      "step": 815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204212302586363e-05,
+      "loss": 1.0508,
+      "step": 816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920182074676503e-05,
+      "loss": 0.9851,
+      "step": 817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919942575206968e-05,
+      "loss": 1.0049,
+      "step": 818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919702731939537e-05,
+      "loss": 0.9961,
+      "step": 819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194625449638427e-05,
+      "loss": 1.0095,
+      "step": 820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9192220143696473e-05,
+      "loss": 1.0232,
+      "step": 821
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918981140246841e-05,
+      "loss": 0.9961,
+      "step": 822
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187399226854427e-05,
+      "loss": 0.9521,
+      "step": 823
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918498361775599e-05,
+      "loss": 0.9958,
+      "step": 824
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918256457607586e-05,
+      "loss": 1.0122,
+      "step": 825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9180142102718064e-05,
+      "loss": 0.9758,
+      "step": 826
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9177716198587926e-05,
+      "loss": 1.0061,
+      "step": 827
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917528686459205e-05,
+      "loss": 1.0076,
+      "step": 828
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917285410163831e-05,
+      "loss": 1.0122,
+      "step": 829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917041791063588e-05,
+      "loss": 1.0386,
+      "step": 830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916797829249519e-05,
+      "loss": 1.0227,
+      "step": 831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9165535248127986e-05,
+      "loss": 0.9834,
+      "step": 832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163088778447263e-05,
+      "loss": 0.9883,
+      "step": 833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9160638884367308e-05,
+      "loss": 1.0315,
+      "step": 834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915818556680369e-05,
+      "loss": 1.012,
+      "step": 835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915572882667325e-05,
+      "loss": 1.0132,
+      "step": 836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915326866489412e-05,
+      "loss": 1.0242,
+      "step": 837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9150805082385694e-05,
+      "loss": 0.9714,
+      "step": 838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9148338080068664e-05,
+      "loss": 1.0264,
+      "step": 839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145867658864983e-05,
+      "loss": 1.0127,
+      "step": 840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143393819697894e-05,
+      "loss": 1.0271,
+      "step": 841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140916563491906e-05,
+      "loss": 0.9812,
+      "step": 842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138435891172812e-05,
+      "loss": 0.9817,
+      "step": 843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135951803667686e-05,
+      "loss": 1.0403,
+      "step": 844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913346430190487e-05,
+      "loss": 0.9905,
+      "step": 845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9130973386813987e-05,
+      "loss": 1.0054,
+      "step": 846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128479059325928e-05,
+      "loss": 1.0042,
+      "step": 847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125981320372866e-05,
+      "loss": 0.9915,
+      "step": 848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123480170888253e-05,
+      "loss": 0.959,
+      "step": 849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9120975611806806e-05,
+      "loss": 1.0088,
+      "step": 850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911846764406452e-05,
+      "loss": 1.0139,
+      "step": 851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115956268598665e-05,
+      "loss": 0.9934,
+      "step": 852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911344148634779e-05,
+      "loss": 1.0376,
+      "step": 853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91109232982517e-05,
+      "loss": 1.0015,
+      "step": 854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9108401705251495e-05,
+      "loss": 0.9895,
+      "step": 855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910587670828953e-05,
+      "loss": 1.0032,
+      "step": 856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910334830830944e-05,
+      "loss": 0.9685,
+      "step": 857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100816506256125e-05,
+      "loss": 1.042,
+      "step": 858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9098281303075776e-05,
+      "loss": 0.9937,
+      "step": 859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909574269971582e-05,
+      "loss": 0.9795,
+      "step": 860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9093200697124997e-05,
+      "loss": 1.0051,
+      "step": 861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090655296253284e-05,
+      "loss": 1.0161,
+      "step": 862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9088106498051938e-05,
+      "loss": 1.0291,
+      "step": 863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085554303473494e-05,
+      "loss": 1.0007,
+      "step": 864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9082998713471743e-05,
+      "loss": 1.0149,
+      "step": 865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908043972900176e-05,
+      "loss": 1.0061,
+      "step": 866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077877351019872e-05,
+      "loss": 1.0315,
+      "step": 867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9075311580483687e-05,
+      "loss": 1.0208,
+      "step": 868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9072742418352076e-05,
+      "loss": 1.0051,
+      "step": 869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070169865585175e-05,
+      "loss": 0.9958,
+      "step": 870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9067593923144395e-05,
+      "loss": 0.9873,
+      "step": 871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9065014591992404e-05,
+      "loss": 0.9873,
+      "step": 872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9062431873093143e-05,
+      "loss": 1.0286,
+      "step": 873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9059845767411816e-05,
+      "loss": 0.9863,
+      "step": 874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057256275914895e-05,
+      "loss": 0.9917,
+      "step": 875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9054663399570118e-05,
+      "loss": 1.0208,
+      "step": 876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905206713934648e-05,
+      "loss": 0.9995,
+      "step": 877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9049467496214256e-05,
+      "loss": 1.0049,
+      "step": 878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904686447114497e-05,
+      "loss": 1.0093,
+      "step": 879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044258065111414e-05,
+      "loss": 0.9885,
+      "step": 880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904164827908765e-05,
+      "loss": 0.96,
+      "step": 881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039035114048996e-05,
+      "loss": 0.9836,
+      "step": 882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903641857097204e-05,
+      "loss": 1.0039,
+      "step": 883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903379865083462e-05,
+      "loss": 0.9565,
+      "step": 884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903117535461585e-05,
+      "loss": 1.0298,
+      "step": 885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90285486832961e-05,
+      "loss": 1.0259,
+      "step": 886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902591863785699e-05,
+      "loss": 0.9819,
+      "step": 887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023285219281428e-05,
+      "loss": 0.9995,
+      "step": 888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9020648428553555e-05,
+      "loss": 1.0122,
+      "step": 889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901800826665879e-05,
+      "loss": 0.9578,
+      "step": 890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015364734583803e-05,
+      "loss": 1.0029,
+      "step": 891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9012717833316524e-05,
+      "loss": 0.9756,
+      "step": 892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901006756384615e-05,
+      "loss": 1.0281,
+      "step": 893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9007413927163124e-05,
+      "loss": 1.021,
+      "step": 894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900475692425916e-05,
+      "loss": 0.991,
+      "step": 895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9002096556127225e-05,
+      "loss": 1.0227,
+      "step": 896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8999432823761534e-05,
+      "loss": 1.0027,
+      "step": 897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996765728157577e-05,
+      "loss": 0.9326,
+      "step": 898
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899409527031209e-05,
+      "loss": 0.9705,
+      "step": 899
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991421451223067e-05,
+      "loss": 1.0107,
+      "step": 900
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8988744271889757e-05,
+      "loss": 0.9951,
+      "step": 901
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898606373331267e-05,
+      "loss": 1.0352,
+      "step": 902
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983379836493565e-05,
+      "loss": 0.9978,
+      "step": 903
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898069258243546e-05,
+      "loss": 0.989,
+      "step": 904
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897800197214263e-05,
+      "loss": 1.0259,
+      "step": 905
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89753080066206e-05,
+      "loss": 1.001,
+      "step": 906
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8972610686876143e-05,
+      "loss": 1.0088,
+      "step": 907
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89699100139173e-05,
+      "loss": 1.0232,
+      "step": 908
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8967205988753357e-05,
+      "loss": 0.9724,
+      "step": 909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896449861239485e-05,
+      "loss": 0.9819,
+      "step": 910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896178788585357e-05,
+      "loss": 0.9749,
+      "step": 911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959073810142564e-05,
+      "loss": 1.0239,
+      "step": 912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956356386276128e-05,
+      "loss": 1.0093,
+      "step": 913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953635615269805e-05,
+      "loss": 1.0505,
+      "step": 914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8950911498140397e-05,
+      "loss": 1.0007,
+      "step": 915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894818403590595e-05,
+      "loss": 1.0571,
+      "step": 916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945453229585758e-05,
+      "loss": 0.9922,
+      "step": 917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8942719080200372e-05,
+      "loss": 1.0435,
+      "step": 918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939981588771595e-05,
+      "loss": 1.0129,
+      "step": 919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8937240756322466e-05,
+      "loss": 1.0115,
+      "step": 920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893449658387728e-05,
+      "loss": 1.0027,
+      "step": 921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8931749072461586e-05,
+      "loss": 1.0464,
+      "step": 922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928998223102165e-05,
+      "loss": 1.0579,
+      "step": 923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926244036827067e-05,
+      "loss": 0.9966,
+      "step": 924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892348651466557e-05,
+      "loss": 0.9998,
+      "step": 925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8920725657648207e-05,
+      "loss": 1.0151,
+      "step": 926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8917961466806756e-05,
+      "loss": 0.9753,
+      "step": 927
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8915193943174245e-05,
+      "loss": 0.9827,
+      "step": 928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8912423087784938e-05,
+      "loss": 1.042,
+      "step": 929
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8909648901674352e-05,
+      "loss": 1.0168,
+      "step": 930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906871385879253e-05,
+      "loss": 0.9758,
+      "step": 931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890409054143764e-05,
+      "loss": 0.9497,
+      "step": 932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8901306369388755e-05,
+      "loss": 1.0149,
+      "step": 933
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898518870773097e-05,
+      "loss": 0.97,
+      "step": 934
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88957280466324e-05,
+      "loss": 0.9771,
+      "step": 935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889293389800964e-05,
+      "loss": 1.0142,
+      "step": 936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889013642594904e-05,
+      "loss": 1.011,
+      "step": 937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8887335631496056e-05,
+      "loss": 0.9998,
+      "step": 938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88845315156974e-05,
+      "loss": 0.9695,
+      "step": 939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881724079601004e-05,
+      "loss": 1.0151,
+      "step": 940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887891332425607e-05,
+      "loss": 0.9463,
+      "step": 941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8876099250713012e-05,
+      "loss": 0.9863,
+      "step": 942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88732818600235e-05,
+      "loss": 1.0061,
+      "step": 943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887046115324044e-05,
+      "loss": 1.0007,
+      "step": 944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8867637131417978e-05,
+      "loss": 0.9956,
+      "step": 945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88648097956115e-05,
+      "loss": 1.0393,
+      "step": 946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8861979146877623e-05,
+      "loss": 1.04,
+      "step": 947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8859145186274214e-05,
+      "loss": 0.9543,
+      "step": 948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8856307914860364e-05,
+      "loss": 0.9597,
+      "step": 949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8853467333696414e-05,
+      "loss": 0.9841,
+      "step": 950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850623443843935e-05,
+      "loss": 1.0288,
+      "step": 951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884777624636574e-05,
+      "loss": 1.0376,
+      "step": 952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844925742325866e-05,
+      "loss": 1.0103,
+      "step": 953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88420719327896e-05,
+      "loss": 0.9534,
+      "step": 954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8839214818823463e-05,
+      "loss": 0.9985,
+      "step": 955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883635440149519e-05,
+      "loss": 0.9834,
+      "step": 956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8833490681873785e-05,
+      "loss": 0.9939,
+      "step": 957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8830623661029457e-05,
+      "loss": 0.9927,
+      "step": 958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882775334003366e-05,
+      "loss": 1.0002,
+      "step": 959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8824879719959085e-05,
+      "loss": 1.0288,
+      "step": 960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882200280187965e-05,
+      "loss": 1.0239,
+      "step": 961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.881912258687051e-05,
+      "loss": 0.96,
+      "step": 962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8816239076008042e-05,
+      "loss": 1.0295,
+      "step": 963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8813352270369868e-05,
+      "loss": 1.0046,
+      "step": 964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8810462171034832e-05,
+      "loss": 0.9692,
+      "step": 965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880756877908302e-05,
+      "loss": 0.9719,
+      "step": 966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880467209559573e-05,
+      "loss": 0.9893,
+      "step": 967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880177212165551e-05,
+      "loss": 0.9343,
+      "step": 968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798868858346126e-05,
+      "loss": 0.9827,
+      "step": 969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8795962306752572e-05,
+      "loss": 1.0056,
+      "step": 970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879305246796108e-05,
+      "loss": 1.0193,
+      "step": 971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8790139343059103e-05,
+      "loss": 0.9844,
+      "step": 972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878722293313532e-05,
+      "loss": 0.9814,
+      "step": 973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8784303239279652e-05,
+      "loss": 0.9973,
+      "step": 974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878138026258323e-05,
+      "loss": 0.9719,
+      "step": 975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877845400413842e-05,
+      "loss": 1.0242,
+      "step": 976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877552446503882e-05,
+      "loss": 0.9915,
+      "step": 977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877259164637924e-05,
+      "loss": 0.9827,
+      "step": 978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769655549255724e-05,
+      "loss": 1.0591,
+      "step": 979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876671617476554e-05,
+      "loss": 1.0222,
+      "step": 980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876377352400719e-05,
+      "loss": 1.0012,
+      "step": 981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876082759808038e-05,
+      "loss": 0.9521,
+      "step": 982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875787839808606e-05,
+      "loss": 0.9839,
+      "step": 983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875492592512639e-05,
+      "loss": 0.981,
+      "step": 984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8751970180304763e-05,
+      "loss": 0.9871,
+      "step": 985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749011164725787e-05,
+      "loss": 1.0337,
+      "step": 986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8746048879495296e-05,
+      "loss": 1.001,
+      "step": 987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743083325720347e-05,
+      "loss": 1.0007,
+      "step": 988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8740114504509214e-05,
+      "loss": 0.9773,
+      "step": 989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8737142416971398e-05,
+      "loss": 1.042,
+      "step": 990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8734167064217617e-05,
+      "loss": 0.9985,
+      "step": 991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8731188447359806e-05,
+      "loss": 1.0222,
+      "step": 992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8728206567511134e-05,
+      "loss": 0.9893,
+      "step": 993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8725221425785965e-05,
+      "loss": 1.032,
+      "step": 994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872223302329991e-05,
+      "loss": 0.9395,
+      "step": 995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8719241361169777e-05,
+      "loss": 0.9846,
+      "step": 996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87162464405136e-05,
+      "loss": 1.0815,
+      "step": 997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8713248262450635e-05,
+      "loss": 1.0203,
+      "step": 998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871024682810135e-05,
+      "loss": 1.0081,
+      "step": 999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8707242138587425e-05,
+      "loss": 0.988,
+      "step": 1000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8704234195031775e-05,
+      "loss": 1.0376,
+      "step": 1001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8701222998558505e-05,
+      "loss": 1.0139,
+      "step": 1002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8698208550292957e-05,
+      "loss": 0.9971,
+      "step": 1003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8695190851361685e-05,
+      "loss": 0.9802,
+      "step": 1004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869216990289244e-05,
+      "loss": 1.0039,
+      "step": 1005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8689145706014217e-05,
+      "loss": 0.9956,
+      "step": 1006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8686118261857193e-05,
+      "loss": 0.9946,
+      "step": 1007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8683087571552786e-05,
+      "loss": 0.9768,
+      "step": 1008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8680053636233612e-05,
+      "loss": 0.979,
+      "step": 1009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8677016457033498e-05,
+      "loss": 0.9751,
+      "step": 1010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86739760350875e-05,
+      "loss": 0.9741,
+      "step": 1011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867093237153186e-05,
+      "loss": 1.021,
+      "step": 1012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866788546750406e-05,
+      "loss": 1.0054,
+      "step": 1013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8664835324142772e-05,
+      "loss": 0.9729,
+      "step": 1014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8661781942587885e-05,
+      "loss": 0.9875,
+      "step": 1015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8658725323980497e-05,
+      "loss": 0.988,
+      "step": 1016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655665469462928e-05,
+      "loss": 0.9392,
+      "step": 1017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865260238017868e-05,
+      "loss": 0.9614,
+      "step": 1018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8649536057272495e-05,
+      "loss": 1.0195,
+      "step": 1019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86464665018903e-05,
+      "loss": 1.0198,
+      "step": 1020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643393715179248e-05,
+      "loss": 0.989,
+      "step": 1021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864031769828768e-05,
+      "loss": 1.0203,
+      "step": 1022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863723845236516e-05,
+      "loss": 1.0066,
+      "step": 1023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8634155978562458e-05,
+      "loss": 1.0012,
+      "step": 1024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863107027803154e-05,
+      "loss": 0.9822,
+      "step": 1025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8627981351925584e-05,
+      "loss": 0.9966,
+      "step": 1026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624889201398978e-05,
+      "loss": 1.0,
+      "step": 1027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8621793827607303e-05,
+      "loss": 0.9932,
+      "step": 1028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861869523170736e-05,
+      "loss": 0.9924,
+      "step": 1029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861559341485714e-05,
+      "loss": 0.97,
+      "step": 1030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861248837821584e-05,
+      "loss": 1.0271,
+      "step": 1031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8609380122943875e-05,
+      "loss": 1.0386,
+      "step": 1032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8606268650202847e-05,
+      "loss": 0.96,
+      "step": 1033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8603153961155563e-05,
+      "loss": 1.0056,
+      "step": 1034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860003605696604e-05,
+      "loss": 0.9756,
+      "step": 1035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8596914938799483e-05,
+      "loss": 0.9783,
+      "step": 1036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593790607822312e-05,
+      "loss": 1.0115,
+      "step": 1037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8590663065202137e-05,
+      "loss": 1.0159,
+      "step": 1038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8587532312107774e-05,
+      "loss": 0.9851,
+      "step": 1039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584398349709244e-05,
+      "loss": 0.9714,
+      "step": 1040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8581261179177753e-05,
+      "loss": 1.0259,
+      "step": 1041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857812080168572e-05,
+      "loss": 0.9934,
+      "step": 1042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857497721840675e-05,
+      "loss": 0.9807,
+      "step": 1043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8571830430515655e-05,
+      "loss": 1.0405,
+      "step": 1044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856868043918845e-05,
+      "loss": 0.9729,
+      "step": 1045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8565527245602327e-05,
+      "loss": 1.012,
+      "step": 1046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8562370850935697e-05,
+      "loss": 1.0244,
+      "step": 1047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855921125636815e-05,
+      "loss": 0.9648,
+      "step": 1048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8556048463080483e-05,
+      "loss": 0.9719,
+      "step": 1049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552882472254686e-05,
+      "loss": 1.0251,
+      "step": 1050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8549713285073934e-05,
+      "loss": 1.0222,
+      "step": 1051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8546540902722618e-05,
+      "loss": 1.0239,
+      "step": 1052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8543365326386305e-05,
+      "loss": 1.0107,
+      "step": 1053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8540186557251755e-05,
+      "loss": 0.9954,
+      "step": 1054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8537004596506937e-05,
+      "loss": 0.9858,
+      "step": 1055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533819445340993e-05,
+      "loss": 0.9658,
+      "step": 1056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8530631104944275e-05,
+      "loss": 0.9707,
+      "step": 1057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8527439576508318e-05,
+      "loss": 0.9702,
+      "step": 1058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8524244861225843e-05,
+      "loss": 1.0137,
+      "step": 1059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8521046960290773e-05,
+      "loss": 1.0649,
+      "step": 1060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851784587489822e-05,
+      "loss": 0.9993,
+      "step": 1061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851464160624448e-05,
+      "loss": 0.9739,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851143415552704e-05,
+      "loss": 1.0493,
+      "step": 1063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850822352394458e-05,
+      "loss": 0.9609,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8505009712696965e-05,
+      "loss": 0.9778,
+      "step": 1065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850179272298525e-05,
+      "loss": 1.019,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849857255601168e-05,
+      "loss": 1.009,
+      "step": 1067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8495349212979685e-05,
+      "loss": 0.9673,
+      "step": 1068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8492122695093884e-05,
+      "loss": 0.9705,
+      "step": 1069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8488893003560072e-05,
+      "loss": 0.9565,
+      "step": 1070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8485660139585245e-05,
+      "loss": 1.0146,
+      "step": 1071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848242410437758e-05,
+      "loss": 0.9763,
+      "step": 1072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8479184899146433e-05,
+      "loss": 1.0186,
+      "step": 1073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847594252510235e-05,
+      "loss": 1.0291,
+      "step": 1074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472696983457066e-05,
+      "loss": 0.9958,
+      "step": 1075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8469448275423484e-05,
+      "loss": 0.9839,
+      "step": 1076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8466196402215708e-05,
+      "loss": 0.9932,
+      "step": 1077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8462941365049016e-05,
+      "loss": 0.9487,
+      "step": 1078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459683165139867e-05,
+      "loss": 0.9939,
+      "step": 1079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8456421803705907e-05,
+      "loss": 0.9751,
+      "step": 1080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845315728196596e-05,
+      "loss": 0.99,
+      "step": 1081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8449889601140036e-05,
+      "loss": 1.0007,
+      "step": 1082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446618762449314e-05,
+      "loss": 1.0178,
+      "step": 1083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844334476711617e-05,
+      "loss": 1.0393,
+      "step": 1084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844006761636414e-05,
+      "loss": 0.9844,
+      "step": 1085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8436787311417962e-05,
+      "loss": 0.9834,
+      "step": 1086
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8433503853503534e-05,
+      "loss": 0.9998,
+      "step": 1087
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843021724384794e-05,
+      "loss": 1.0059,
+      "step": 1088
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8426927483679444e-05,
+      "loss": 0.9695,
+      "step": 1089
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8423634574227474e-05,
+      "loss": 1.0137,
+      "step": 1090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842033851672266e-05,
+      "loss": 0.9519,
+      "step": 1091
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8417039312396783e-05,
+      "loss": 0.9419,
+      "step": 1092
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413736962482814e-05,
+      "loss": 0.998,
+      "step": 1093
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8410431468214898e-05,
+      "loss": 1.0107,
+      "step": 1094
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840712283082835e-05,
+      "loss": 0.9995,
+      "step": 1095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8403811051559666e-05,
+      "loss": 0.9922,
+      "step": 1096
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840049613164651e-05,
+      "loss": 0.9814,
+      "step": 1097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8397178072327727e-05,
+      "loss": 1.0005,
+      "step": 1098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839385687484333e-05,
+      "loss": 0.9429,
+      "step": 1099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8390532540434502e-05,
+      "loss": 1.0076,
+      "step": 1100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8387205070343602e-05,
+      "loss": 1.0181,
+      "step": 1101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383874465814173e-05,
+      "loss": 0.999,
+      "step": 1102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.83805407280909e-05,
+      "loss": 0.9729,
+      "step": 1103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.837720385841967e-05,
+      "loss": 1.0186,
+      "step": 1104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8373863858047522e-05,
+      "loss": 0.9612,
+      "step": 1105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837052072822267e-05,
+      "loss": 0.9927,
+      "step": 1106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83671744701945e-05,
+      "loss": 0.9534,
+      "step": 1107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8363825085213554e-05,
+      "loss": 0.9902,
+      "step": 1108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836047257453157e-05,
+      "loss": 1.0159,
+      "step": 1109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8357116939401424e-05,
+      "loss": 0.9673,
+      "step": 1110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353758181077178e-05,
+      "loss": 0.991,
+      "step": 1111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835039630081405e-05,
+      "loss": 0.9932,
+      "step": 1112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347031299868444e-05,
+      "loss": 1.0117,
+      "step": 1113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8343663179497904e-05,
+      "loss": 0.9839,
+      "step": 1114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834029194096116e-05,
+      "loss": 0.9631,
+      "step": 1115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8336917585518095e-05,
+      "loss": 0.9827,
+      "step": 1116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8333540114429766e-05,
+      "loss": 0.9912,
+      "step": 1117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8330159528958388e-05,
+      "loss": 0.9827,
+      "step": 1118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8326775830367347e-05,
+      "loss": 1.0442,
+      "step": 1119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832338901992118e-05,
+      "loss": 0.9646,
+      "step": 1120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8319999098885602e-05,
+      "loss": 0.9863,
+      "step": 1121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831660606852748e-05,
+      "loss": 1.0183,
+      "step": 1122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8313209930114842e-05,
+      "loss": 0.9641,
+      "step": 1123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830981068491689e-05,
+      "loss": 0.9685,
+      "step": 1124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8306408334203968e-05,
+      "loss": 0.999,
+      "step": 1125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.83030028792476e-05,
+      "loss": 1.0017,
+      "step": 1126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8299594321320458e-05,
+      "loss": 1.0144,
+      "step": 1127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8296182661696382e-05,
+      "loss": 1.019,
+      "step": 1128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829276790165036e-05,
+      "loss": 0.9675,
+      "step": 1129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8289350042458544e-05,
+      "loss": 0.9868,
+      "step": 1130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8285929085398248e-05,
+      "loss": 0.9729,
+      "step": 1131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828250503174794e-05,
+      "loss": 0.9709,
+      "step": 1132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8279077882787248e-05,
+      "loss": 1.0188,
+      "step": 1133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8275647639796955e-05,
+      "loss": 0.958,
+      "step": 1134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8272214304058996e-05,
+      "loss": 0.9976,
+      "step": 1135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8268777876856466e-05,
+      "loss": 1.0164,
+      "step": 1136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826533835947362e-05,
+      "loss": 0.9587,
+      "step": 1137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826189575319586e-05,
+      "loss": 0.9629,
+      "step": 1138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8258450059309746e-05,
+      "loss": 1.0005,
+      "step": 1139
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8255001279102994e-05,
+      "loss": 0.9983,
+      "step": 1140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8251549413864462e-05,
+      "loss": 0.9749,
+      "step": 1141
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824809446488418e-05,
+      "loss": 1.0042,
+      "step": 1142
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8244636433453315e-05,
+      "loss": 1.0295,
+      "step": 1143
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8241175320864187e-05,
+      "loss": 0.9771,
+      "step": 1144
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8237711128410283e-05,
+      "loss": 0.9873,
+      "step": 1145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823424385738622e-05,
+      "loss": 0.9907,
+      "step": 1146
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823077350908778e-05,
+      "loss": 0.99,
+      "step": 1147
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8227300084811882e-05,
+      "loss": 0.991,
+      "step": 1148
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223823585856616e-05,
+      "loss": 0.9827,
+      "step": 1149
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8220344013521193e-05,
+      "loss": 1.0083,
+      "step": 1150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8216861369106e-05,
+      "loss": 1.0007,
+      "step": 1151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821337565391255e-05,
+      "loss": 0.9854,
+      "step": 1152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820988686924352e-05,
+      "loss": 1.0146,
+      "step": 1153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8206395016402724e-05,
+      "loss": 1.0115,
+      "step": 1154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8202900096695125e-05,
+      "loss": 0.9709,
+      "step": 1155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8199402111426828e-05,
+      "loss": 1.0159,
+      "step": 1156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.81959010619051e-05,
+      "loss": 0.9744,
+      "step": 1157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8192396949438337e-05,
+      "loss": 0.9585,
+      "step": 1158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818888977533608e-05,
+      "loss": 0.9568,
+      "step": 1159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818537954090902e-05,
+      "loss": 0.9807,
+      "step": 1160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181866247468997e-05,
+      "loss": 1.0032,
+      "step": 1161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8178349896328976e-05,
+      "loss": 1.0112,
+      "step": 1162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8174830488803083e-05,
+      "loss": 0.9724,
+      "step": 1163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817130802620658e-05,
+      "loss": 0.9756,
+      "step": 1164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816778250985587e-05,
+      "loss": 1.033,
+      "step": 1165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816425394106849e-05,
+      "loss": 0.9634,
+      "step": 1166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8160722321163137e-05,
+      "loss": 1.0364,
+      "step": 1167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815718765145963e-05,
+      "loss": 0.9961,
+      "step": 1168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153649933278932e-05,
+      "loss": 0.9127,
+      "step": 1169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815010916794315e-05,
+      "loss": 1.0566,
+      "step": 1170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8146565356775527e-05,
+      "loss": 0.998,
+      "step": 1171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814301850110044e-05,
+      "loss": 1.0242,
+      "step": 1172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.813946860224342e-05,
+      "loss": 1.0269,
+      "step": 1173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.813591566153111e-05,
+      "loss": 0.9426,
+      "step": 1174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8132359680291314e-05,
+      "loss": 0.9521,
+      "step": 1175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8128800659852952e-05,
+      "loss": 1.0205,
+      "step": 1176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8125238601546097e-05,
+      "loss": 0.9939,
+      "step": 1177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8121673506701945e-05,
+      "loss": 0.9194,
+      "step": 1178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8118105376652833e-05,
+      "loss": 0.9941,
+      "step": 1179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811453421273223e-05,
+      "loss": 0.9441,
+      "step": 1180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8110960016274742e-05,
+      "loss": 0.9824,
+      "step": 1181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.81073827886161e-05,
+      "loss": 1.0486,
+      "step": 1182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8103802531093185e-05,
+      "loss": 0.9851,
+      "step": 1183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8100219245043987e-05,
+      "loss": 1.0012,
+      "step": 1184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8096632931807647e-05,
+      "loss": 1.0105,
+      "step": 1185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8093043592724426e-05,
+      "loss": 0.9922,
+      "step": 1186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808945122913572e-05,
+      "loss": 0.979,
+      "step": 1187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808585584238406e-05,
+      "loss": 1.002,
+      "step": 1188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8082257433813098e-05,
+      "loss": 0.9277,
+      "step": 1189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8078656004767623e-05,
+      "loss": 0.978,
+      "step": 1190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8075051556593544e-05,
+      "loss": 1.0237,
+      "step": 1191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8071444090637907e-05,
+      "loss": 1.019,
+      "step": 1192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067833608248878e-05,
+      "loss": 0.9832,
+      "step": 1193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8064220110775762e-05,
+      "loss": 1.0051,
+      "step": 1194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8060603599568978e-05,
+      "loss": 0.9827,
+      "step": 1195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8056984075980075e-05,
+      "loss": 1.0112,
+      "step": 1196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8053361541361734e-05,
+      "loss": 1.0156,
+      "step": 1197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8049735997067754e-05,
+      "loss": 0.9705,
+      "step": 1198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8046107444453064e-05,
+      "loss": 0.978,
+      "step": 1199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804247588487371e-05,
+      "loss": 0.9719,
+      "step": 1200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8038841319686873e-05,
+      "loss": 0.9832,
+      "step": 1201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8035203750250845e-05,
+      "loss": 0.989,
+      "step": 1202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8031563177925048e-05,
+      "loss": 0.999,
+      "step": 1203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8027919604070026e-05,
+      "loss": 0.9978,
+      "step": 1204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8024273030047442e-05,
+      "loss": 0.9648,
+      "step": 1205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8020623457220082e-05,
+      "loss": 0.9729,
+      "step": 1206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8016970886951853e-05,
+      "loss": 0.9827,
+      "step": 1207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801331532060778e-05,
+      "loss": 0.9497,
+      "step": 1208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8009656759554008e-05,
+      "loss": 0.9788,
+      "step": 1209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8005995205157807e-05,
+      "loss": 1.0063,
+      "step": 1210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800233065878756e-05,
+      "loss": 0.9954,
+      "step": 1211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7998663121812763e-05,
+      "loss": 1.0122,
+      "step": 1212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799499259560404e-05,
+      "loss": 1.0066,
+      "step": 1213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7991319081533132e-05,
+      "loss": 1.0359,
+      "step": 1214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7987642580972884e-05,
+      "loss": 1.0107,
+      "step": 1215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7983963095297273e-05,
+      "loss": 1.0205,
+      "step": 1216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7980280625881383e-05,
+      "loss": 1.0083,
+      "step": 1217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797659517410141e-05,
+      "loss": 0.9465,
+      "step": 1218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7972906741334672e-05,
+      "loss": 0.9719,
+      "step": 1219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7969215328959596e-05,
+      "loss": 0.9937,
+      "step": 1220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796552093835573e-05,
+      "loss": 0.9802,
+      "step": 1221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7961823570903724e-05,
+      "loss": 1.012,
+      "step": 1222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7958123227985345e-05,
+      "loss": 1.0283,
+      "step": 1223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7954419910983478e-05,
+      "loss": 0.9382,
+      "step": 1224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795071362128211e-05,
+      "loss": 0.9692,
+      "step": 1225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7947004360266345e-05,
+      "loss": 1.0303,
+      "step": 1226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7943292129322395e-05,
+      "loss": 1.0144,
+      "step": 1227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7939576929837585e-05,
+      "loss": 1.0105,
+      "step": 1228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935858763200345e-05,
+      "loss": 0.938,
+      "step": 1229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7932137630800217e-05,
+      "loss": 0.9854,
+      "step": 1230
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7928413534027847e-05,
+      "loss": 0.9788,
+      "step": 1231
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7924686474274997e-05,
+      "loss": 1.0,
+      "step": 1232
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7920956452934527e-05,
+      "loss": 0.978,
+      "step": 1233
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7917223471400413e-05,
+      "loss": 0.9885,
+      "step": 1234
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7913487531067728e-05,
+      "loss": 0.9907,
+      "step": 1235
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790974863333266e-05,
+      "loss": 1.0103,
+      "step": 1236
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7906006779592493e-05,
+      "loss": 0.9873,
+      "step": 1237
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7902261971245623e-05,
+      "loss": 1.0134,
+      "step": 1238
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7898514209691544e-05,
+      "loss": 1.0271,
+      "step": 1239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789476349633086e-05,
+      "loss": 0.9905,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7891009832565272e-05,
+      "loss": 0.9902,
+      "step": 1241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7887253219797595e-05,
+      "loss": 0.9905,
+      "step": 1242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788349365943173e-05,
+      "loss": 0.9595,
+      "step": 1243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7879731152872692e-05,
+      "loss": 1.0266,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7875965701526586e-05,
+      "loss": 1.0098,
+      "step": 1245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787219730680063e-05,
+      "loss": 0.9565,
+      "step": 1246
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7868425970103133e-05,
+      "loss": 0.9851,
+      "step": 1247
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786465169284351e-05,
+      "loss": 0.988,
+      "step": 1248
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7860874476432268e-05,
+      "loss": 1.0208,
+      "step": 1249
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7857094322281016e-05,
+      "loss": 0.9729,
+      "step": 1250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7853311231802462e-05,
+      "loss": 0.9919,
+      "step": 1251
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784952520641041e-05,
+      "loss": 0.9868,
+      "step": 1252
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7845736247519764e-05,
+      "loss": 1.0132,
+      "step": 1253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7841944356546515e-05,
+      "loss": 0.95,
+      "step": 1254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783814953490776e-05,
+      "loss": 1.0127,
+      "step": 1255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783435178402168e-05,
+      "loss": 0.9902,
+      "step": 1256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7830551105307572e-05,
+      "loss": 0.9514,
+      "step": 1257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7826747500185798e-05,
+      "loss": 0.9829,
+      "step": 1258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782294097007784e-05,
+      "loss": 0.9888,
+      "step": 1259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781913151640626e-05,
+      "loss": 0.98,
+      "step": 1260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7815319140594707e-05,
+      "loss": 0.9766,
+      "step": 1261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7811503844067934e-05,
+      "loss": 0.9758,
+      "step": 1262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7807685628251782e-05,
+      "loss": 0.9753,
+      "step": 1263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7803864494573186e-05,
+      "loss": 1.0046,
+      "step": 1264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7800040444460158e-05,
+      "loss": 0.9954,
+      "step": 1265
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7796213479341816e-05,
+      "loss": 0.9575,
+      "step": 1266
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779238360064836e-05,
+      "loss": 1.0039,
+      "step": 1267
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7788550809811075e-05,
+      "loss": 1.0376,
+      "step": 1268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7784715108262344e-05,
+      "loss": 1.0129,
+      "step": 1269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778087649743563e-05,
+      "loss": 1.0181,
+      "step": 1270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7777034978765492e-05,
+      "loss": 1.0007,
+      "step": 1271
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7773190553687562e-05,
+      "loss": 0.9998,
+      "step": 1272
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776934322363857e-05,
+      "loss": 0.9827,
+      "step": 1273
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7765492990056323e-05,
+      "loss": 0.9832,
+      "step": 1274
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776163985437972e-05,
+      "loss": 0.9824,
+      "step": 1275
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7757783818048745e-05,
+      "loss": 1.0161,
+      "step": 1276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775392488250446e-05,
+      "loss": 0.96,
+      "step": 1277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7750063049189016e-05,
+      "loss": 0.9858,
+      "step": 1278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7746198319545643e-05,
+      "loss": 0.9788,
+      "step": 1279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7742330695018655e-05,
+      "loss": 0.99,
+      "step": 1280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7738460177053447e-05,
+      "loss": 1.0132,
+      "step": 1281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7734586767096496e-05,
+      "loss": 0.9509,
+      "step": 1282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7730710466595363e-05,
+      "loss": 0.9805,
+      "step": 1283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7726831276998682e-05,
+      "loss": 1.0142,
+      "step": 1284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7722949199756173e-05,
+      "loss": 0.9792,
+      "step": 1285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7719064236318634e-05,
+      "loss": 0.9573,
+      "step": 1286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771517638813794e-05,
+      "loss": 0.9805,
+      "step": 1287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771128565666705e-05,
+      "loss": 1.0081,
+      "step": 1288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770739204335998e-05,
+      "loss": 0.9468,
+      "step": 1289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7703495549671854e-05,
+      "loss": 0.9702,
+      "step": 1290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769959617705885e-05,
+      "loss": 0.9839,
+      "step": 1291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769569392697823e-05,
+      "loss": 0.9333,
+      "step": 1292
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7691788800888326e-05,
+      "loss": 0.9883,
+      "step": 1293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768788080024856e-05,
+      "loss": 0.989,
+      "step": 1294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7683969926519404e-05,
+      "loss": 1.0012,
+      "step": 1295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7680056181162424e-05,
+      "loss": 0.9888,
+      "step": 1296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767613956564025e-05,
+      "loss": 0.9866,
+      "step": 1297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7672220081416585e-05,
+      "loss": 1.0129,
+      "step": 1298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766829772995621e-05,
+      "loss": 1.0115,
+      "step": 1299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7664372512724967e-05,
+      "loss": 0.9426,
+      "step": 1300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 1.0334,
+      "step": 1301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765651348681864e-05,
+      "loss": 0.9961,
+      "step": 1302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7652579681080602e-05,
+      "loss": 0.988,
+      "step": 1303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7648643015445798e-05,
+      "loss": 1.02,
+      "step": 1304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7644703491385428e-05,
+      "loss": 0.9668,
+      "step": 1305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764076111037175e-05,
+      "loss": 0.9932,
+      "step": 1306
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7636815873878103e-05,
+      "loss": 1.0625,
+      "step": 1307
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7632867783378887e-05,
+      "loss": 1.0225,
+      "step": 1308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762891684034957e-05,
+      "loss": 0.9863,
+      "step": 1309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7624963046266685e-05,
+      "loss": 0.9871,
+      "step": 1310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7621006402607827e-05,
+      "loss": 0.9917,
+      "step": 1311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7617046910851666e-05,
+      "loss": 0.9929,
+      "step": 1312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7613084572477926e-05,
+      "loss": 0.9663,
+      "step": 1313
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76091193889674e-05,
+      "loss": 1.0081,
+      "step": 1314
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7605151361801946e-05,
+      "loss": 0.9668,
+      "step": 1315
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7601180492464477e-05,
+      "loss": 1.0027,
+      "step": 1316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7597206782438978e-05,
+      "loss": 1.0061,
+      "step": 1317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7593230233210486e-05,
+      "loss": 1.0276,
+      "step": 1318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758925084626511e-05,
+      "loss": 1.0024,
+      "step": 1319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7585268623090004e-05,
+      "loss": 0.9509,
+      "step": 1320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75812835651734e-05,
+      "loss": 0.9583,
+      "step": 1321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757729567400458e-05,
+      "loss": 1.0068,
+      "step": 1322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7573304951073876e-05,
+      "loss": 1.0203,
+      "step": 1323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75693113978727e-05,
+      "loss": 0.9993,
+      "step": 1324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7565315015893506e-05,
+      "loss": 1.0044,
+      "step": 1325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7561315806629803e-05,
+      "loss": 0.9375,
+      "step": 1326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755731377157617e-05,
+      "loss": 0.9785,
+      "step": 1327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7553308912228228e-05,
+      "loss": 0.9524,
+      "step": 1328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7549301230082664e-05,
+      "loss": 0.9885,
+      "step": 1329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754529072663721e-05,
+      "loss": 0.9548,
+      "step": 1330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541277403390667e-05,
+      "loss": 1.0173,
+      "step": 1331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7537261261842874e-05,
+      "loss": 0.9756,
+      "step": 1332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7533242303494736e-05,
+      "loss": 0.9465,
+      "step": 1333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7529220529848195e-05,
+      "loss": 1.0339,
+      "step": 1334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752519594240626e-05,
+      "loss": 0.9736,
+      "step": 1335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7521168542672993e-05,
+      "loss": 0.9824,
+      "step": 1336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7517138332153493e-05,
+      "loss": 1.0515,
+      "step": 1337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751310531235392e-05,
+      "loss": 0.979,
+      "step": 1338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750906948478148e-05,
+      "loss": 1.0361,
+      "step": 1339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7505030850944425e-05,
+      "loss": 1.032,
+      "step": 1340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750098941235206e-05,
+      "loss": 0.9927,
+      "step": 1341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749694517051475e-05,
+      "loss": 0.9949,
+      "step": 1342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7492898126943887e-05,
+      "loss": 0.9897,
+      "step": 1343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748884828315192e-05,
+      "loss": 0.9612,
+      "step": 1344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748479564065234e-05,
+      "loss": 1.0378,
+      "step": 1345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7480740200959684e-05,
+      "loss": 0.9641,
+      "step": 1346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747668196558955e-05,
+      "loss": 1.0144,
+      "step": 1347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747262093605856e-05,
+      "loss": 0.9561,
+      "step": 1348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746855711388439e-05,
+      "loss": 0.9841,
+      "step": 1349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7464490500585756e-05,
+      "loss": 0.9634,
+      "step": 1350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7460421097682425e-05,
+      "loss": 0.9727,
+      "step": 1351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7456348906695196e-05,
+      "loss": 1.0073,
+      "step": 1352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7452273929145915e-05,
+      "loss": 0.9487,
+      "step": 1353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7448196166557468e-05,
+      "loss": 1.0171,
+      "step": 1354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744411562045379e-05,
+      "loss": 0.9083,
+      "step": 1355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7440032292359842e-05,
+      "loss": 0.9951,
+      "step": 1356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743594618380163e-05,
+      "loss": 0.9802,
+      "step": 1357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743185729630621e-05,
+      "loss": 1.0234,
+      "step": 1358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7427765631401663e-05,
+      "loss": 1.0032,
+      "step": 1359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742367119061711e-05,
+      "loss": 0.9763,
+      "step": 1360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7419573975482713e-05,
+      "loss": 0.9966,
+      "step": 1361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741547398752967e-05,
+      "loss": 0.9841,
+      "step": 1362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741137122829022e-05,
+      "loss": 0.9924,
+      "step": 1363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740726569929762e-05,
+      "loss": 0.9832,
+      "step": 1364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7403157402086193e-05,
+      "loss": 0.9917,
+      "step": 1365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739904633819126e-05,
+      "loss": 0.988,
+      "step": 1366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7394932509149203e-05,
+      "loss": 0.9634,
+      "step": 1367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7390815916497432e-05,
+      "loss": 0.9753,
+      "step": 1368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.738669656177438e-05,
+      "loss": 0.967,
+      "step": 1369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738257444651952e-05,
+      "loss": 0.9709,
+      "step": 1370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7378449572273362e-05,
+      "loss": 0.9573,
+      "step": 1371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7374321940577428e-05,
+      "loss": 0.947,
+      "step": 1372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7370191552974293e-05,
+      "loss": 1.0032,
+      "step": 1373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7366058411007543e-05,
+      "loss": 0.9553,
+      "step": 1374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736192251622181e-05,
+      "loss": 0.9624,
+      "step": 1375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735778387016275e-05,
+      "loss": 1.0032,
+      "step": 1376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7353642474377032e-05,
+      "loss": 0.9944,
+      "step": 1377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7349498330412372e-05,
+      "loss": 0.9924,
+      "step": 1378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7345351439817505e-05,
+      "loss": 1.0227,
+      "step": 1379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7341201804142193e-05,
+      "loss": 0.9739,
+      "step": 1380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7337049424937227e-05,
+      "loss": 0.9749,
+      "step": 1381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7332894303754412e-05,
+      "loss": 1.0251,
+      "step": 1382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7328736442146595e-05,
+      "loss": 0.9336,
+      "step": 1383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7324575841667636e-05,
+      "loss": 0.9978,
+      "step": 1384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732041250387242e-05,
+      "loss": 0.9795,
+      "step": 1385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7316246430316855e-05,
+      "loss": 0.9622,
+      "step": 1386
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731207762255787e-05,
+      "loss": 0.9717,
+      "step": 1387
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7307906082153425e-05,
+      "loss": 1.0127,
+      "step": 1388
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730373181066249e-05,
+      "loss": 0.9727,
+      "step": 1389
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729955480964506e-05,
+      "loss": 1.0068,
+      "step": 1390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729537508066215e-05,
+      "loss": 1.0239,
+      "step": 1391
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729119262527579e-05,
+      "loss": 0.9866,
+      "step": 1392
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7287007445049045e-05,
+      "loss": 1.0107,
+      "step": 1393
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7282819541545977e-05,
+      "loss": 0.9854,
+      "step": 1394
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7278628916331678e-05,
+      "loss": 0.9788,
+      "step": 1395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7274435570972256e-05,
+      "loss": 0.9949,
+      "step": 1396
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7270239507034833e-05,
+      "loss": 1.0247,
+      "step": 1397
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726604072608755e-05,
+      "loss": 0.9985,
+      "step": 1398
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7261839229699555e-05,
+      "loss": 0.9978,
+      "step": 1399
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7257635019441027e-05,
+      "loss": 1.0293,
+      "step": 1400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725342809688314e-05,
+      "loss": 0.9814,
+      "step": 1401
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72492184635981e-05,
+      "loss": 1.0166,
+      "step": 1402
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724500612115911e-05,
+      "loss": 0.9658,
+      "step": 1403
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7240791071140398e-05,
+      "loss": 1.0332,
+      "step": 1404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7236573315117195e-05,
+      "loss": 0.9988,
+      "step": 1405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7232352854665747e-05,
+      "loss": 0.9966,
+      "step": 1406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7228129691363312e-05,
+      "loss": 0.9429,
+      "step": 1407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7223903826788156e-05,
+      "loss": 1.0017,
+      "step": 1408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7219675262519555e-05,
+      "loss": 1.0293,
+      "step": 1409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7215444000137797e-05,
+      "loss": 0.9617,
+      "step": 1410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7211210041224168e-05,
+      "loss": 0.9487,
+      "step": 1411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720697338736098e-05,
+      "loss": 1.0051,
+      "step": 1412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720273404013153e-05,
+      "loss": 1.011,
+      "step": 1413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7198492001120143e-05,
+      "loss": 1.0193,
+      "step": 1414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7194247271912132e-05,
+      "loss": 1.0132,
+      "step": 1415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7189999854093824e-05,
+      "loss": 1.0046,
+      "step": 1416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7185749749252556e-05,
+      "loss": 1.0151,
+      "step": 1417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7181496958976656e-05,
+      "loss": 1.0154,
+      "step": 1418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7177241484855467e-05,
+      "loss": 0.9988,
+      "step": 1419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717298332847933e-05,
+      "loss": 1.0142,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7168722491439586e-05,
+      "loss": 0.9673,
+      "step": 1421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7164458975328586e-05,
+      "loss": 0.9961,
+      "step": 1422
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7160192781739674e-05,
+      "loss": 1.0051,
+      "step": 1423
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7155923912267196e-05,
+      "loss": 0.9775,
+      "step": 1424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7151652368506508e-05,
+      "loss": 0.9917,
+      "step": 1425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7147378152053945e-05,
+      "loss": 0.9805,
+      "step": 1426
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7143101264506863e-05,
+      "loss": 0.9929,
+      "step": 1427
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7138821707463598e-05,
+      "loss": 0.9724,
+      "step": 1428
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7134539482523503e-05,
+      "loss": 0.9802,
+      "step": 1429
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713025459128691e-05,
+      "loss": 1.0232,
+      "step": 1430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712596703535515e-05,
+      "loss": 0.948,
+      "step": 1431
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7121676816330558e-05,
+      "loss": 1.0171,
+      "step": 1432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711738393581647e-05,
+      "loss": 0.9832,
+      "step": 1433
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7113088395417193e-05,
+      "loss": 0.9839,
+      "step": 1434
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7108790196738046e-05,
+      "loss": 0.9851,
+      "step": 1435
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7104489341385342e-05,
+      "loss": 0.9429,
+      "step": 1436
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710018583096638e-05,
+      "loss": 1.0037,
+      "step": 1437
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7095879667089454e-05,
+      "loss": 0.9888,
+      "step": 1438
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709157085136385e-05,
+      "loss": 0.9482,
+      "step": 1439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708725938539984e-05,
+      "loss": 0.9666,
+      "step": 1440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7082945270808697e-05,
+      "loss": 1.0149,
+      "step": 1441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707862850920267e-05,
+      "loss": 1.0164,
+      "step": 1442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7074309102195018e-05,
+      "loss": 0.9919,
+      "step": 1443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706998705139996e-05,
+      "loss": 1.0327,
+      "step": 1444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7065662358432726e-05,
+      "loss": 1.0181,
+      "step": 1445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706133502490953e-05,
+      "loss": 0.9824,
+      "step": 1446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7057005052447558e-05,
+      "loss": 0.9585,
+      "step": 1447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7052672442665006e-05,
+      "loss": 0.9958,
+      "step": 1448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704833719718103e-05,
+      "loss": 0.9668,
+      "step": 1449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043999317615792e-05,
+      "loss": 0.9641,
+      "step": 1450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7039658805590428e-05,
+      "loss": 1.0154,
+      "step": 1451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7035315662727055e-05,
+      "loss": 0.978,
+      "step": 1452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703096989064879e-05,
+      "loss": 0.9412,
+      "step": 1453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702662149097971e-05,
+      "loss": 0.9983,
+      "step": 1454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7022270465344884e-05,
+      "loss": 0.9885,
+      "step": 1455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7017916815370367e-05,
+      "loss": 0.9792,
+      "step": 1456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701356054268319e-05,
+      "loss": 0.9863,
+      "step": 1457
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7009201648911367e-05,
+      "loss": 0.9055,
+      "step": 1458
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7004840135683887e-05,
+      "loss": 0.9927,
+      "step": 1459
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700047600463072e-05,
+      "loss": 1.0095,
+      "step": 1460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6996109257382812e-05,
+      "loss": 1.001,
+      "step": 1461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6991739895572094e-05,
+      "loss": 1.0032,
+      "step": 1462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698736792083147e-05,
+      "loss": 1.0359,
+      "step": 1463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6982993334794818e-05,
+      "loss": 0.9895,
+      "step": 1464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697861613909699e-05,
+      "loss": 1.0012,
+      "step": 1465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6974236335373825e-05,
+      "loss": 0.9858,
+      "step": 1466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696985392526212e-05,
+      "loss": 0.9351,
+      "step": 1467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6965468910399663e-05,
+      "loss": 1.0071,
+      "step": 1468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69610812924252e-05,
+      "loss": 0.9839,
+      "step": 1469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6956691072978463e-05,
+      "loss": 1.0173,
+      "step": 1470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695229825370015e-05,
+      "loss": 0.9268,
+      "step": 1471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6947902836231925e-05,
+      "loss": 0.9817,
+      "step": 1472
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6943504822216433e-05,
+      "loss": 0.981,
+      "step": 1473
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6939104213297285e-05,
+      "loss": 0.9851,
+      "step": 1474
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6934701011119064e-05,
+      "loss": 0.9639,
+      "step": 1475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6930295217327315e-05,
+      "loss": 0.9653,
+      "step": 1476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6925886833568564e-05,
+      "loss": 0.946,
+      "step": 1477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6921475861490287e-05,
+      "loss": 0.9583,
+      "step": 1478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691706230274095e-05,
+      "loss": 0.9766,
+      "step": 1479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6912646158969965e-05,
+      "loss": 1.0032,
+      "step": 1480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6908227431827723e-05,
+      "loss": 1.0337,
+      "step": 1481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6903806122965574e-05,
+      "loss": 0.9934,
+      "step": 1482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6899382234035834e-05,
+      "loss": 1.0288,
+      "step": 1483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6894955766691788e-05,
+      "loss": 0.9841,
+      "step": 1484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6890526722587678e-05,
+      "loss": 0.9617,
+      "step": 1485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688609510337871e-05,
+      "loss": 0.9971,
+      "step": 1486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688166091072106e-05,
+      "loss": 0.9976,
+      "step": 1487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687722414627185e-05,
+      "loss": 0.9941,
+      "step": 1488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687278481168918e-05,
+      "loss": 1.0046,
+      "step": 1489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6868342908632108e-05,
+      "loss": 0.9939,
+      "step": 1490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6863898438760638e-05,
+      "loss": 0.9934,
+      "step": 1491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6859451403735746e-05,
+      "loss": 1.0066,
+      "step": 1492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6855001805219366e-05,
+      "loss": 0.9832,
+      "step": 1493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685054964487438e-05,
+      "loss": 0.9551,
+      "step": 1494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684609492436464e-05,
+      "loss": 0.9604,
+      "step": 1495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6841637645354954e-05,
+      "loss": 0.9663,
+      "step": 1496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683717780951107e-05,
+      "loss": 0.9702,
+      "step": 1497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6832715418499718e-05,
+      "loss": 1.0256,
+      "step": 1498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682825047398855e-05,
+      "loss": 1.0095,
+      "step": 1499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6823782977646204e-05,
+      "loss": 1.0134,
+      "step": 1500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6819312931142255e-05,
+      "loss": 0.9949,
+      "step": 1501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681484033614723e-05,
+      "loss": 0.9963,
+      "step": 1502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6810365194332616e-05,
+      "loss": 1.022,
+      "step": 1503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6805887507370847e-05,
+      "loss": 0.9573,
+      "step": 1504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680140727693531e-05,
+      "loss": 1.01,
+      "step": 1505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679692450470034e-05,
+      "loss": 1.02,
+      "step": 1506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6792439192341228e-05,
+      "loss": 0.9697,
+      "step": 1507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6787951341534206e-05,
+      "loss": 0.9543,
+      "step": 1508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6783460953956457e-05,
+      "loss": 0.9277,
+      "step": 1509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6778968031286123e-05,
+      "loss": 1.0032,
+      "step": 1510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6774472575202273e-05,
+      "loss": 0.9912,
+      "step": 1511
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6769974587384944e-05,
+      "loss": 0.9895,
+      "step": 1512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6765474069515102e-05,
+      "loss": 0.9597,
+      "step": 1513
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6760971023274665e-05,
+      "loss": 0.9233,
+      "step": 1514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6756465450346505e-05,
+      "loss": 0.989,
+      "step": 1515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675195735241442e-05,
+      "loss": 0.9766,
+      "step": 1516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6747446731163168e-05,
+      "loss": 0.9226,
+      "step": 1517
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674293358827844e-05,
+      "loss": 0.9543,
+      "step": 1518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6738417925446875e-05,
+      "loss": 0.9646,
+      "step": 1519
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673389974435605e-05,
+      "loss": 0.9849,
+      "step": 1520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6729379046694486e-05,
+      "loss": 0.9973,
+      "step": 1521
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6724855834151643e-05,
+      "loss": 0.9739,
+      "step": 1522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672033010841792e-05,
+      "loss": 0.9836,
+      "step": 1523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6715801871184655e-05,
+      "loss": 0.9343,
+      "step": 1524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6711271124144128e-05,
+      "loss": 0.9927,
+      "step": 1525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6706737868989554e-05,
+      "loss": 0.9521,
+      "step": 1526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670220210741509e-05,
+      "loss": 1.0098,
+      "step": 1527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6697663841115822e-05,
+      "loss": 0.9683,
+      "step": 1528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6693123071787774e-05,
+      "loss": 1.0444,
+      "step": 1529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668857980112791e-05,
+      "loss": 0.9343,
+      "step": 1530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6684034030834125e-05,
+      "loss": 1.0288,
+      "step": 1531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6679485762605247e-05,
+      "loss": 0.9558,
+      "step": 1532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6674934998141048e-05,
+      "loss": 1.0002,
+      "step": 1533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6670381739142217e-05,
+      "loss": 0.9729,
+      "step": 1534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6665825987310386e-05,
+      "loss": 0.9741,
+      "step": 1535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6661267744348114e-05,
+      "loss": 0.9514,
+      "step": 1536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6656707011958894e-05,
+      "loss": 0.9778,
+      "step": 1537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6652143791847145e-05,
+      "loss": 1.0266,
+      "step": 1538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6647578085718223e-05,
+      "loss": 0.9548,
+      "step": 1539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6643009895278403e-05,
+      "loss": 0.9546,
+      "step": 1540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6638439222234903e-05,
+      "loss": 0.9585,
+      "step": 1541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6633866068295847e-05,
+      "loss": 0.9883,
+      "step": 1542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6629290435170308e-05,
+      "loss": 0.9536,
+      "step": 1543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6624712324568277e-05,
+      "loss": 0.9692,
+      "step": 1544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6620131738200664e-05,
+      "loss": 0.9758,
+      "step": 1545
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6615548677779317e-05,
+      "loss": 0.9958,
+      "step": 1546
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6610963145016997e-05,
+      "loss": 0.9399,
+      "step": 1547
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6606375141627394e-05,
+      "loss": 1.0244,
+      "step": 1548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660178466932513e-05,
+      "loss": 0.9399,
+      "step": 1549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6597191729825732e-05,
+      "loss": 0.9187,
+      "step": 1550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6592596324845663e-05,
+      "loss": 0.9608,
+      "step": 1551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65879984561023e-05,
+      "loss": 0.9834,
+      "step": 1552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6583398125313946e-05,
+      "loss": 0.9656,
+      "step": 1553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6578795334199818e-05,
+      "loss": 0.9543,
+      "step": 1554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6574190084480058e-05,
+      "loss": 0.9739,
+      "step": 1555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6569582377875725e-05,
+      "loss": 1.0095,
+      "step": 1556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656497221610879e-05,
+      "loss": 1.0071,
+      "step": 1557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6560359600902157e-05,
+      "loss": 0.9553,
+      "step": 1558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655574453397963e-05,
+      "loss": 1.0269,
+      "step": 1559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6551127017065944e-05,
+      "loss": 0.9998,
+      "step": 1560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6546507051886735e-05,
+      "loss": 0.9795,
+      "step": 1561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6541884640168563e-05,
+      "loss": 0.9297,
+      "step": 1562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65372597836389e-05,
+      "loss": 0.9763,
+      "step": 1563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6532632484026135e-05,
+      "loss": 0.9927,
+      "step": 1564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6528002743059562e-05,
+      "loss": 0.9736,
+      "step": 1565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65233705624694e-05,
+      "loss": 1.0142,
+      "step": 1566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6518735943986765e-05,
+      "loss": 0.9736,
+      "step": 1567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6514098889343696e-05,
+      "loss": 0.9736,
+      "step": 1568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650945940027313e-05,
+      "loss": 0.9675,
+      "step": 1569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6504817478508934e-05,
+      "loss": 0.9609,
+      "step": 1570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6500173125785864e-05,
+      "loss": 0.9922,
+      "step": 1571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649552634383959e-05,
+      "loss": 0.9475,
+      "step": 1572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64908771344067e-05,
+      "loss": 0.979,
+      "step": 1573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648622549922467e-05,
+      "loss": 0.9766,
+      "step": 1574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6481571440031907e-05,
+      "loss": 1.0208,
+      "step": 1575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64769149585677e-05,
+      "loss": 0.9932,
+      "step": 1576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6472256056572254e-05,
+      "loss": 0.9805,
+      "step": 1577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646759473578669e-05,
+      "loss": 0.9319,
+      "step": 1578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646293099795301e-05,
+      "loss": 0.9819,
+      "step": 1579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645826484481413e-05,
+      "loss": 1.0061,
+      "step": 1580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6453596278113877e-05,
+      "loss": 0.9685,
+      "step": 1581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6448925299596966e-05,
+      "loss": 1.0063,
+      "step": 1582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6444251911009024e-05,
+      "loss": 0.9854,
+      "step": 1583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6439576114096573e-05,
+      "loss": 0.9944,
+      "step": 1584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6434897910607035e-05,
+      "loss": 0.9749,
+      "step": 1585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6430217302288736e-05,
+      "loss": 1.0105,
+      "step": 1586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642553429089089e-05,
+      "loss": 0.9517,
+      "step": 1587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6420848878163626e-05,
+      "loss": 0.9604,
+      "step": 1588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6416161065857957e-05,
+      "loss": 0.9629,
+      "step": 1589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6411470855725794e-05,
+      "loss": 0.9839,
+      "step": 1590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640677824951995e-05,
+      "loss": 0.9937,
+      "step": 1591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640208324899413e-05,
+      "loss": 1.0203,
+      "step": 1592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6397385855902933e-05,
+      "loss": 0.9797,
+      "step": 1593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6392686072001854e-05,
+      "loss": 0.9907,
+      "step": 1594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638798389904728e-05,
+      "loss": 0.9644,
+      "step": 1595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638327933879649e-05,
+      "loss": 0.9868,
+      "step": 1596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6378572393007657e-05,
+      "loss": 0.9761,
+      "step": 1597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6373863063439852e-05,
+      "loss": 0.9604,
+      "step": 1598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6369151351853022e-05,
+      "loss": 1.0298,
+      "step": 1599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636443726000802e-05,
+      "loss": 0.9485,
+      "step": 1600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635972078966657e-05,
+      "loss": 0.9478,
+      "step": 1601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6355001942591302e-05,
+      "loss": 0.9817,
+      "step": 1602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635028072054573e-05,
+      "loss": 0.9492,
+      "step": 1603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6345557125294253e-05,
+      "loss": 1.0083,
+      "step": 1604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634083115860215e-05,
+      "loss": 1.0056,
+      "step": 1605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6336102822235604e-05,
+      "loss": 0.99,
+      "step": 1606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6331372117961665e-05,
+      "loss": 0.9644,
+      "step": 1607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632663904754828e-05,
+      "loss": 0.9529,
+      "step": 1608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6321903612764276e-05,
+      "loss": 0.9817,
+      "step": 1609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6317165815379364e-05,
+      "loss": 0.9841,
+      "step": 1610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6312425657164137e-05,
+      "loss": 0.979,
+      "step": 1611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6307683139890073e-05,
+      "loss": 1.0029,
+      "step": 1612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630293826532953e-05,
+      "loss": 0.9861,
+      "step": 1613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6298191035255745e-05,
+      "loss": 1.0176,
+      "step": 1614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629344145144284e-05,
+      "loss": 1.0203,
+      "step": 1615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.628868951566581e-05,
+      "loss": 0.9482,
+      "step": 1616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628393522970053e-05,
+      "loss": 0.9597,
+      "step": 1617
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6279178595323766e-05,
+      "loss": 0.9958,
+      "step": 1618
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6274419614313145e-05,
+      "loss": 0.968,
+      "step": 1619
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626965828844718e-05,
+      "loss": 0.9707,
+      "step": 1620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6264894619505255e-05,
+      "loss": 0.9155,
+      "step": 1621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626012860926763e-05,
+      "loss": 0.9709,
+      "step": 1622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6255360259515453e-05,
+      "loss": 0.9939,
+      "step": 1623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625058957203073e-05,
+      "loss": 0.9885,
+      "step": 1624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624581654859634e-05,
+      "loss": 1.0024,
+      "step": 1625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6241041190996058e-05,
+      "loss": 0.9155,
+      "step": 1626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62362635010145e-05,
+      "loss": 0.9561,
+      "step": 1627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623148348043717e-05,
+      "loss": 1.0127,
+      "step": 1628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622670113105045e-05,
+      "loss": 1.02,
+      "step": 1629
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6221916454641572e-05,
+      "loss": 0.9514,
+      "step": 1630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6217129452998662e-05,
+      "loss": 0.9524,
+      "step": 1631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621234012791069e-05,
+      "loss": 0.9946,
+      "step": 1632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620754848116752e-05,
+      "loss": 0.9727,
+      "step": 1633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620275451455986e-05,
+      "loss": 0.9917,
+      "step": 1634
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61979582298793e-05,
+      "loss": 0.9795,
+      "step": 1635
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6193159628918288e-05,
+      "loss": 0.959,
+      "step": 1636
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6188358713470143e-05,
+      "loss": 1.0068,
+      "step": 1637
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618355548532905e-05,
+      "loss": 0.968,
+      "step": 1638
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6178749946290056e-05,
+      "loss": 1.0068,
+      "step": 1639
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6173942098149068e-05,
+      "loss": 0.9919,
+      "step": 1640
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6169131942702854e-05,
+      "loss": 1.011,
+      "step": 1641
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6164319481749063e-05,
+      "loss": 0.9702,
+      "step": 1642
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6159504717086178e-05,
+      "loss": 1.0112,
+      "step": 1643
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6154687650513566e-05,
+      "loss": 0.9929,
+      "step": 1644
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6149868283831442e-05,
+      "loss": 0.999,
+      "step": 1645
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6145046618840883e-05,
+      "loss": 0.9907,
+      "step": 1646
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614022265734383e-05,
+      "loss": 0.9976,
+      "step": 1647
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613539640114307e-05,
+      "loss": 0.9304,
+      "step": 1648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6130567852042264e-05,
+      "loss": 0.9766,
+      "step": 1649
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6125737011845917e-05,
+      "loss": 1.009,
+      "step": 1650
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6120903882359393e-05,
+      "loss": 0.9934,
+      "step": 1651
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6116068465388913e-05,
+      "loss": 0.9363,
+      "step": 1652
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6111230762741563e-05,
+      "loss": 0.9792,
+      "step": 1653
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6106390776225257e-05,
+      "loss": 0.9856,
+      "step": 1654
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6101548507648793e-05,
+      "loss": 0.9705,
+      "step": 1655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6096703958821804e-05,
+      "loss": 0.9788,
+      "step": 1656
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609185713155477e-05,
+      "loss": 0.9434,
+      "step": 1657
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608700802765904e-05,
+      "loss": 0.9438,
+      "step": 1658
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6082156648946808e-05,
+      "loss": 0.9995,
+      "step": 1659
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6077302997231104e-05,
+      "loss": 0.9436,
+      "step": 1660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607244707432583e-05,
+      "loss": 0.9202,
+      "step": 1661
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6067588882045714e-05,
+      "loss": 0.9429,
+      "step": 1662
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606272842220636e-05,
+      "loss": 1.0203,
+      "step": 1663
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6057865696624184e-05,
+      "loss": 0.9548,
+      "step": 1664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605300070711648e-05,
+      "loss": 0.95,
+      "step": 1665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6048133455501373e-05,
+      "loss": 0.9631,
+      "step": 1666
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6043263943597832e-05,
+      "loss": 0.9617,
+      "step": 1667
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6038392173225678e-05,
+      "loss": 0.9697,
+      "step": 1668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603351814620558e-05,
+      "loss": 0.9617,
+      "step": 1669
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602864186435903e-05,
+      "loss": 0.9619,
+      "step": 1670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6023763329508386e-05,
+      "loss": 0.9517,
+      "step": 1671
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6018882543476835e-05,
+      "loss": 0.9634,
+      "step": 1672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6013999508088404e-05,
+      "loss": 0.9597,
+      "step": 1673
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6009114225167972e-05,
+      "loss": 0.9939,
+      "step": 1674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6004226696541247e-05,
+      "loss": 0.9792,
+      "step": 1675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5999336924034777e-05,
+      "loss": 0.9678,
+      "step": 1676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599444490947596e-05,
+      "loss": 0.9602,
+      "step": 1677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5989550654693012e-05,
+      "loss": 1.0022,
+      "step": 1678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598465416151501e-05,
+      "loss": 0.9766,
+      "step": 1679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597975543177185e-05,
+      "loss": 1.009,
+      "step": 1680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5974854467294263e-05,
+      "loss": 0.9846,
+      "step": 1681
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596995126991383e-05,
+      "loss": 0.9963,
+      "step": 1682
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965045841462954e-05,
+      "loss": 0.958,
+      "step": 1683
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5960138183774874e-05,
+      "loss": 0.9673,
+      "step": 1684
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595522829868367e-05,
+      "loss": 0.9958,
+      "step": 1685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5950316188024237e-05,
+      "loss": 1.0391,
+      "step": 1686
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5945401853632324e-05,
+      "loss": 1.0051,
+      "step": 1687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5940485297344496e-05,
+      "loss": 0.9595,
+      "step": 1688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5935566520998145e-05,
+      "loss": 0.9929,
+      "step": 1689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5930645526431515e-05,
+      "loss": 0.9753,
+      "step": 1690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5925722315483648e-05,
+      "loss": 0.9822,
+      "step": 1691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592079688999444e-05,
+      "loss": 0.9351,
+      "step": 1692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59158692518046e-05,
+      "loss": 1.0017,
+      "step": 1693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5910939402755666e-05,
+      "loss": 0.917,
+      "step": 1694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5906007344690018e-05,
+      "loss": 0.9424,
+      "step": 1695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5901073079450835e-05,
+      "loss": 0.9587,
+      "step": 1696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589613660888214e-05,
+      "loss": 0.95,
+      "step": 1697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5891197934828773e-05,
+      "loss": 1.011,
+      "step": 1698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5886257059136398e-05,
+      "loss": 0.9714,
+      "step": 1699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5881313983651507e-05,
+      "loss": 0.9907,
+      "step": 1700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5876368710221404e-05,
+      "loss": 0.9204,
+      "step": 1701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5871421240694223e-05,
+      "loss": 0.9868,
+      "step": 1702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586647157691892e-05,
+      "loss": 0.9507,
+      "step": 1703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5861519720745257e-05,
+      "loss": 1.0063,
+      "step": 1704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5856565674023833e-05,
+      "loss": 0.9685,
+      "step": 1705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5851609438606058e-05,
+      "loss": 0.97,
+      "step": 1706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584665101634416e-05,
+      "loss": 0.9705,
+      "step": 1707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5841690409091178e-05,
+      "loss": 0.9644,
+      "step": 1708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5836727618700984e-05,
+      "loss": 0.9626,
+      "step": 1709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583176264702824e-05,
+      "loss": 0.9673,
+      "step": 1710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582679549592846e-05,
+      "loss": 0.9373,
+      "step": 1711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582182616725793e-05,
+      "loss": 0.99,
+      "step": 1712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5816854662873783e-05,
+      "loss": 0.9802,
+      "step": 1713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5811880984633953e-05,
+      "loss": 0.9795,
+      "step": 1714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5806905134397186e-05,
+      "loss": 0.969,
+      "step": 1715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5801927114023038e-05,
+      "loss": 0.9849,
+      "step": 1716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5796946925371877e-05,
+      "loss": 0.927,
+      "step": 1717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5791964570304887e-05,
+      "loss": 0.9929,
+      "step": 1718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5786980050684055e-05,
+      "loss": 0.9839,
+      "step": 1719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5781993368372176e-05,
+      "loss": 0.9839,
+      "step": 1720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577700452523286e-05,
+      "loss": 0.9507,
+      "step": 1721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5772013523130526e-05,
+      "loss": 0.9355,
+      "step": 1722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5767020363930384e-05,
+      "loss": 0.9595,
+      "step": 1723
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576202504949847e-05,
+      "loss": 1.0071,
+      "step": 1724
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575702758170161e-05,
+      "loss": 1.02,
+      "step": 1725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5752027962407445e-05,
+      "loss": 0.9995,
+      "step": 1726
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5747026193484417e-05,
+      "loss": 0.9583,
+      "step": 1727
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5742022276801768e-05,
+      "loss": 0.9236,
+      "step": 1728
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5737016214229543e-05,
+      "loss": 0.96,
+      "step": 1729
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5732008007638594e-05,
+      "loss": 0.9683,
+      "step": 1730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5726997658900572e-05,
+      "loss": 0.9734,
+      "step": 1731
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572198516988793e-05,
+      "loss": 0.9629,
+      "step": 1732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5716970542473913e-05,
+      "loss": 0.9929,
+      "step": 1733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5711953778532576e-05,
+      "loss": 1.0085,
+      "step": 1734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5706934879938764e-05,
+      "loss": 0.9663,
+      "step": 1735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5701913848568125e-05,
+      "loss": 0.9495,
+      "step": 1736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5696890686297103e-05,
+      "loss": 1.0117,
+      "step": 1737
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5691865395002938e-05,
+      "loss": 1.0076,
+      "step": 1738
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568683797656366e-05,
+      "loss": 0.9563,
+      "step": 1739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5681808432858107e-05,
+      "loss": 0.9468,
+      "step": 1740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5676776765765892e-05,
+      "loss": 0.9436,
+      "step": 1741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5671742977167446e-05,
+      "loss": 0.9075,
+      "step": 1742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5666707068943975e-05,
+      "loss": 1.0107,
+      "step": 1743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5661669042977477e-05,
+      "loss": 0.9326,
+      "step": 1744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5656628901150758e-05,
+      "loss": 1.0015,
+      "step": 1745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565158664534739e-05,
+      "loss": 0.928,
+      "step": 1746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5646542277451752e-05,
+      "loss": 0.9343,
+      "step": 1747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5641495799349016e-05,
+      "loss": 0.9922,
+      "step": 1748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5636447212925126e-05,
+      "loss": 0.9944,
+      "step": 1749
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5631396520066825e-05,
+      "loss": 0.9604,
+      "step": 1750
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5626343722661647e-05,
+      "loss": 0.9858,
+      "step": 1751
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5621288822597903e-05,
+      "loss": 0.9688,
+      "step": 1752
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5616231821764687e-05,
+      "loss": 0.9958,
+      "step": 1753
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5611172722051895e-05,
+      "loss": 0.957,
+      "step": 1754
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5606111525350193e-05,
+      "loss": 0.9517,
+      "step": 1755
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560104823355103e-05,
+      "loss": 0.9563,
+      "step": 1756
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.559598284854665e-05,
+      "loss": 0.989,
+      "step": 1757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.559091537223007e-05,
+      "loss": 0.9749,
+      "step": 1758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5585845806495084e-05,
+      "loss": 0.9661,
+      "step": 1759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5580774153236284e-05,
+      "loss": 0.9756,
+      "step": 1760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5575700414349023e-05,
+      "loss": 0.958,
+      "step": 1761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557062459172945e-05,
+      "loss": 0.9438,
+      "step": 1762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5565546687274478e-05,
+      "loss": 0.9768,
+      "step": 1763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5560466702881805e-05,
+      "loss": 1.0015,
+      "step": 1764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555538464044991e-05,
+      "loss": 1.0122,
+      "step": 1765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555030050187804e-05,
+      "loss": 0.9868,
+      "step": 1766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554521428906622e-05,
+      "loss": 0.8577,
+      "step": 1767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5540126003915262e-05,
+      "loss": 0.8059,
+      "step": 1768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5535035648326735e-05,
+      "loss": 0.8054,
+      "step": 1769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5529943224202993e-05,
+      "loss": 0.8225,
+      "step": 1770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5524848733447163e-05,
+      "loss": 0.8025,
+      "step": 1771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5519752177963135e-05,
+      "loss": 0.7817,
+      "step": 1772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5514653559655584e-05,
+      "loss": 0.812,
+      "step": 1773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5509552880429942e-05,
+      "loss": 0.8203,
+      "step": 1774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.550445014219242e-05,
+      "loss": 0.803,
+      "step": 1775
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5499345346850003e-05,
+      "loss": 0.7341,
+      "step": 1776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5494238496310426e-05,
+      "loss": 0.7893,
+      "step": 1777
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5489129592482213e-05,
+      "loss": 0.7588,
+      "step": 1778
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5484018637274645e-05,
+      "loss": 0.7773,
+      "step": 1779
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547890563259777e-05,
+      "loss": 0.76,
+      "step": 1780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54737905803624e-05,
+      "loss": 0.8081,
+      "step": 1781
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546867348248012e-05,
+      "loss": 0.7668,
+      "step": 1782
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546355434086328e-05,
+      "loss": 0.7832,
+      "step": 1783
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5458433157424978e-05,
+      "loss": 0.7922,
+      "step": 1784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5453309934079092e-05,
+      "loss": 0.7668,
+      "step": 1785
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5448184672740257e-05,
+      "loss": 0.802,
+      "step": 1786
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5443057375323864e-05,
+      "loss": 0.8093,
+      "step": 1787
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543792804374607e-05,
+      "loss": 0.7756,
+      "step": 1788
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5432796679923794e-05,
+      "loss": 0.7817,
+      "step": 1789
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5427663285774714e-05,
+      "loss": 0.7561,
+      "step": 1790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5422527863217264e-05,
+      "loss": 0.7756,
+      "step": 1791
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5417390414170636e-05,
+      "loss": 0.7988,
+      "step": 1792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5412250940554775e-05,
+      "loss": 0.7839,
+      "step": 1793
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5407109444290402e-05,
+      "loss": 0.7712,
+      "step": 1794
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540196592729897e-05,
+      "loss": 0.7649,
+      "step": 1795
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5396820391502693e-05,
+      "loss": 0.8262,
+      "step": 1796
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5391672838824555e-05,
+      "loss": 0.7593,
+      "step": 1797
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5386523271188276e-05,
+      "loss": 0.7693,
+      "step": 1798
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5381371690518337e-05,
+      "loss": 0.7866,
+      "step": 1799
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537621809873997e-05,
+      "loss": 0.8289,
+      "step": 1800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5371062497779154e-05,
+      "loss": 0.7727,
+      "step": 1801
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5365904889562626e-05,
+      "loss": 0.7451,
+      "step": 1802
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5360745276017873e-05,
+      "loss": 0.7791,
+      "step": 1803
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5355583659073127e-05,
+      "loss": 0.7451,
+      "step": 1804
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535042004065737e-05,
+      "loss": 0.7593,
+      "step": 1805
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5345254422700328e-05,
+      "loss": 0.7891,
+      "step": 1806
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534008680713249e-05,
+      "loss": 0.7949,
+      "step": 1807
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5334917195885067e-05,
+      "loss": 0.7937,
+      "step": 1808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5329745590890038e-05,
+      "loss": 0.7808,
+      "step": 1809
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5324571994080113e-05,
+      "loss": 0.8186,
+      "step": 1810
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5319396407388756e-05,
+      "loss": 0.7698,
+      "step": 1811
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531421883275017e-05,
+      "loss": 0.8262,
+      "step": 1812
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5309039272099294e-05,
+      "loss": 0.7788,
+      "step": 1813
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5303857727371824e-05,
+      "loss": 0.7903,
+      "step": 1814
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5298674200504183e-05,
+      "loss": 0.7649,
+      "step": 1815
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529348869343355e-05,
+      "loss": 0.7881,
+      "step": 1816
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5288301208097828e-05,
+      "loss": 0.7615,
+      "step": 1817
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5283111746435672e-05,
+      "loss": 0.7578,
+      "step": 1818
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5277920310386468e-05,
+      "loss": 0.782,
+      "step": 1819
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5272726901890343e-05,
+      "loss": 0.7637,
+      "step": 1820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526753152288816e-05,
+      "loss": 0.7661,
+      "step": 1821
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5262334175321516e-05,
+      "loss": 0.7615,
+      "step": 1822
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5257134861132754e-05,
+      "loss": 0.7554,
+      "step": 1823
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5251933582264938e-05,
+      "loss": 0.7966,
+      "step": 1824
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5246730340661872e-05,
+      "loss": 0.7676,
+      "step": 1825
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5241525138268101e-05,
+      "loss": 0.822,
+      "step": 1826
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5236317977028891e-05,
+      "loss": 0.7856,
+      "step": 1827
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5231108858890244e-05,
+      "loss": 0.7563,
+      "step": 1828
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.52258977857989e-05,
+      "loss": 0.7424,
+      "step": 1829
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5220684759702318e-05,
+      "loss": 0.7896,
+      "step": 1830
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5215469782548695e-05,
+      "loss": 0.8027,
+      "step": 1831
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5210252856286958e-05,
+      "loss": 0.8267,
+      "step": 1832
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5205033982866752e-05,
+      "loss": 0.759,
+      "step": 1833
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5199813164238469e-05,
+      "loss": 0.8525,
+      "step": 1834
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5194590402353204e-05,
+      "loss": 0.8008,
+      "step": 1835
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5189365699162795e-05,
+      "loss": 0.8013,
+      "step": 1836
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5184139056619805e-05,
+      "loss": 0.7903,
+      "step": 1837
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5178910476677509e-05,
+      "loss": 0.7979,
+      "step": 1838
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517367996128992e-05,
+      "loss": 0.7517,
+      "step": 1839
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5168447512411769e-05,
+      "loss": 0.769,
+      "step": 1840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5163213131998507e-05,
+      "loss": 0.7529,
+      "step": 1841
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5157976822006312e-05,
+      "loss": 0.7627,
+      "step": 1842
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5152738584392083e-05,
+      "loss": 0.8044,
+      "step": 1843
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5147498421113429e-05,
+      "loss": 0.7556,
+      "step": 1844
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5142256334128694e-05,
+      "loss": 0.8037,
+      "step": 1845
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5137012325396926e-05,
+      "loss": 0.7693,
+      "step": 1846
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5131766396877904e-05,
+      "loss": 0.8022,
+      "step": 1847
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.512651855053212e-05,
+      "loss": 0.7537,
+      "step": 1848
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5121268788320778e-05,
+      "loss": 0.7542,
+      "step": 1849
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5116017112205803e-05,
+      "loss": 0.7981,
+      "step": 1850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5110763524149836e-05,
+      "loss": 0.8506,
+      "step": 1851
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5105508026116226e-05,
+      "loss": 0.762,
+      "step": 1852
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5100250620069048e-05,
+      "loss": 0.7917,
+      "step": 1853
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5094991307973073e-05,
+      "loss": 0.7898,
+      "step": 1854
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5089730091793797e-05,
+      "loss": 0.7517,
+      "step": 1855
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508446697349743e-05,
+      "loss": 0.8115,
+      "step": 1856
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5079201955050882e-05,
+      "loss": 0.8323,
+      "step": 1857
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507393503842178e-05,
+      "loss": 0.771,
+      "step": 1858
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5068666225578456e-05,
+      "loss": 0.7742,
+      "step": 1859
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5063395518489957e-05,
+      "loss": 0.7822,
+      "step": 1860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5058122919126028e-05,
+      "loss": 0.7679,
+      "step": 1861
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5052848429457135e-05,
+      "loss": 0.8157,
+      "step": 1862
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.504757205145444e-05,
+      "loss": 0.782,
+      "step": 1863
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5042293787089812e-05,
+      "loss": 0.7561,
+      "step": 1864
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5037013638335827e-05,
+      "loss": 0.8032,
+      "step": 1865
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5031731607165766e-05,
+      "loss": 0.8005,
+      "step": 1866
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5026447695553615e-05,
+      "loss": 0.7576,
+      "step": 1867
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5021161905474051e-05,
+      "loss": 0.7798,
+      "step": 1868
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5015874238902468e-05,
+      "loss": 0.7808,
+      "step": 1869
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5010584697814956e-05,
+      "loss": 0.793,
+      "step": 1870
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5005293284188305e-05,
+      "loss": 0.7725,
+      "step": 1871
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.781,
+      "step": 1872
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4994704847228236e-05,
+      "loss": 0.7561,
+      "step": 1873
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4989407827851896e-05,
+      "loss": 0.7534,
+      "step": 1874
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4984108943850564e-05,
+      "loss": 0.7605,
+      "step": 1875
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4978808197204523e-05,
+      "loss": 0.7898,
+      "step": 1876
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4973505589894753e-05,
+      "loss": 0.7653,
+      "step": 1877
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4968201123902919e-05,
+      "loss": 0.7639,
+      "step": 1878
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4962894801211398e-05,
+      "loss": 0.7583,
+      "step": 1879
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4957586623803246e-05,
+      "loss": 0.8433,
+      "step": 1880
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4952276593662218e-05,
+      "loss": 0.791,
+      "step": 1881
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4946964712772763e-05,
+      "loss": 0.7893,
+      "step": 1882
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4941650983120015e-05,
+      "loss": 0.7903,
+      "step": 1883
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4936335406689805e-05,
+      "loss": 0.8198,
+      "step": 1884
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.493101798546866e-05,
+      "loss": 0.7832,
+      "step": 1885
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4925698721443778e-05,
+      "loss": 0.7981,
+      "step": 1886
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4920377616603064e-05,
+      "loss": 0.7578,
+      "step": 1887
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4915054672935103e-05,
+      "loss": 0.8047,
+      "step": 1888
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4909729892429163e-05,
+      "loss": 0.7871,
+      "step": 1889
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490440327707521e-05,
+      "loss": 0.7871,
+      "step": 1890
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4899074828863884e-05,
+      "loss": 0.781,
+      "step": 1891
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4893744549786516e-05,
+      "loss": 0.7937,
+      "step": 1892
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4888412441835122e-05,
+      "loss": 0.7754,
+      "step": 1893
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.48830785070024e-05,
+      "loss": 0.7954,
+      "step": 1894
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4877742747281729e-05,
+      "loss": 0.7817,
+      "step": 1895
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4872405164667175e-05,
+      "loss": 0.7646,
+      "step": 1896
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4867065761153474e-05,
+      "loss": 0.7573,
+      "step": 1897
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4861724538736056e-05,
+      "loss": 0.7981,
+      "step": 1898
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4856381499411024e-05,
+      "loss": 0.8147,
+      "step": 1899
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485103664517516e-05,
+      "loss": 0.8303,
+      "step": 1900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4845689978025928e-05,
+      "loss": 0.8101,
+      "step": 1901
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4840341499961464e-05,
+      "loss": 0.8054,
+      "step": 1902
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4834991212980585e-05,
+      "loss": 0.7869,
+      "step": 1903
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4829639119082781e-05,
+      "loss": 0.7812,
+      "step": 1904
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482428522026822e-05,
+      "loss": 0.8169,
+      "step": 1905
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4818929518537742e-05,
+      "loss": 0.7637,
+      "step": 1906
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4813572015892862e-05,
+      "loss": 0.8303,
+      "step": 1907
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4808212714335772e-05,
+      "loss": 0.7834,
+      "step": 1908
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.480285161586933e-05,
+      "loss": 0.824,
+      "step": 1909
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4797488722497067e-05,
+      "loss": 0.7771,
+      "step": 1910
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4792124036223187e-05,
+      "loss": 0.7548,
+      "step": 1911
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4786757559052565e-05,
+      "loss": 0.8242,
+      "step": 1912
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4781389292990738e-05,
+      "loss": 0.8,
+      "step": 1913
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4776019240043917e-05,
+      "loss": 0.7644,
+      "step": 1914
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4770647402218985e-05,
+      "loss": 0.7742,
+      "step": 1915
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4765273781523484e-05,
+      "loss": 0.7778,
+      "step": 1916
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4759898379965628e-05,
+      "loss": 0.8066,
+      "step": 1917
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4754521199554291e-05,
+      "loss": 0.7585,
+      "step": 1918
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.474914224229902e-05,
+      "loss": 0.7566,
+      "step": 1919
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4743761510210016e-05,
+      "loss": 0.7832,
+      "step": 1920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473837900529815e-05,
+      "loss": 0.7847,
+      "step": 1921
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473299472957496e-05,
+      "loss": 0.7754,
+      "step": 1922
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.472760868505263e-05,
+      "loss": 0.7944,
+      "step": 1923
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4722220873744017e-05,
+      "loss": 0.8115,
+      "step": 1924
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4716831297662642e-05,
+      "loss": 0.7601,
+      "step": 1925
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4711439958822675e-05,
+      "loss": 0.7861,
+      "step": 1926
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.470604685923895e-05,
+      "loss": 0.7939,
+      "step": 1927
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4700652000926956e-05,
+      "loss": 0.8044,
+      "step": 1928
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4695255385902844e-05,
+      "loss": 0.8008,
+      "step": 1929
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4689857016183422e-05,
+      "loss": 0.7576,
+      "step": 1930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4684456893786148e-05,
+      "loss": 0.804,
+      "step": 1931
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4679055020729134e-05,
+      "loss": 0.7869,
+      "step": 1932
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4673651399031157e-05,
+      "loss": 0.759,
+      "step": 1933
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4668246030711636e-05,
+      "loss": 0.7832,
+      "step": 1934
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4662838917790651e-05,
+      "loss": 0.7495,
+      "step": 1935
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4657430062288924e-05,
+      "loss": 0.7856,
+      "step": 1936
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4652019466227842e-05,
+      "loss": 0.79,
+      "step": 1937
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4646607131629434e-05,
+      "loss": 0.821,
+      "step": 1938
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4641193060516374e-05,
+      "loss": 0.7371,
+      "step": 1939
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4635777254911996e-05,
+      "loss": 0.7688,
+      "step": 1940
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4630359716840276e-05,
+      "loss": 0.8042,
+      "step": 1941
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4624940448325835e-05,
+      "loss": 0.8113,
+      "step": 1942
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4619519451393949e-05,
+      "loss": 0.7561,
+      "step": 1943
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4614096728070533e-05,
+      "loss": 0.8079,
+      "step": 1944
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4608672280382147e-05,
+      "loss": 0.8062,
+      "step": 1945
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4603246110356005e-05,
+      "loss": 0.7493,
+      "step": 1946
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4597818220019947e-05,
+      "loss": 0.7839,
+      "step": 1947
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4592388611402472e-05,
+      "loss": 0.8235,
+      "step": 1948
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.458695728653272e-05,
+      "loss": 0.8162,
+      "step": 1949
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4581524247440455e-05,
+      "loss": 0.7822,
+      "step": 1950
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4576089496156106e-05,
+      "loss": 0.7939,
+      "step": 1951
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4570653034710726e-05,
+      "loss": 0.7678,
+      "step": 1952
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456521486513601e-05,
+      "loss": 0.7708,
+      "step": 1953
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4559774989464297e-05,
+      "loss": 0.7986,
+      "step": 1954
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4554333409728557e-05,
+      "loss": 0.7942,
+      "step": 1955
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4548890127962395e-05,
+      "loss": 0.8198,
+      "step": 1956
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4543445146200064e-05,
+      "loss": 0.8315,
+      "step": 1957
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4537998466476438e-05,
+      "loss": 0.8091,
+      "step": 1958
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4532550090827033e-05,
+      "loss": 0.793,
+      "step": 1959
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4527100021288005e-05,
+      "loss": 0.7842,
+      "step": 1960
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4521648259896127e-05,
+      "loss": 0.7847,
+      "step": 1961
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.451619480868882e-05,
+      "loss": 0.7334,
+      "step": 1962
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4510739669704123e-05,
+      "loss": 0.7698,
+      "step": 1963
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4505282844980717e-05,
+      "loss": 0.7883,
+      "step": 1964
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4499824336557908e-05,
+      "loss": 0.7915,
+      "step": 1965
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449436414647563e-05,
+      "loss": 0.7861,
+      "step": 1966
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4488902276774453e-05,
+      "loss": 0.8105,
+      "step": 1967
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4483438729495562e-05,
+      "loss": 0.7942,
+      "step": 1968
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4477973506680778e-05,
+      "loss": 0.7898,
+      "step": 1969
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4472506610372547e-05,
+      "loss": 0.771,
+      "step": 1970
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4467038042613939e-05,
+      "loss": 0.7627,
+      "step": 1971
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4461567805448646e-05,
+      "loss": 0.8059,
+      "step": 1972
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4456095900920992e-05,
+      "loss": 0.7844,
+      "step": 1973
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4450622331075914e-05,
+      "loss": 0.7905,
+      "step": 1974
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4445147097958983e-05,
+      "loss": 0.7893,
+      "step": 1975
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4439670203616381e-05,
+      "loss": 0.7676,
+      "step": 1976
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443419165009491e-05,
+      "loss": 0.8035,
+      "step": 1977
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4428711439442009e-05,
+      "loss": 0.7834,
+      "step": 1978
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4423229573705712e-05,
+      "loss": 0.7001,
+      "step": 1979
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441774605493469e-05,
+      "loss": 0.7683,
+      "step": 1980
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4412260885178224e-05,
+      "loss": 0.8042,
+      "step": 1981
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4406774066486213e-05,
+      "loss": 0.7841,
+      "step": 1982
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4401285600909175e-05,
+      "loss": 0.8086,
+      "step": 1983
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4395795490498242e-05,
+      "loss": 0.7412,
+      "step": 1984
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4390303737305153e-05,
+      "loss": 0.7617,
+      "step": 1985
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4384810343382277e-05,
+      "loss": 0.7913,
+      "step": 1986
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4379315310782578e-05,
+      "loss": 0.7856,
+      "step": 1987
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4373818641559646e-05,
+      "loss": 0.7712,
+      "step": 1988
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4368320337767683e-05,
+      "loss": 0.7654,
+      "step": 1989
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4362820401461488e-05,
+      "loss": 0.8149,
+      "step": 1990
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4357318834696483e-05,
+      "loss": 0.8276,
+      "step": 1991
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.43518156395287e-05,
+      "loss": 0.7969,
+      "step": 1992
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4346310818014763e-05,
+      "loss": 0.7703,
+      "step": 1993
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4340804372211928e-05,
+      "loss": 0.7839,
+      "step": 1994
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4335296304178034e-05,
+      "loss": 0.7822,
+      "step": 1995
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4329786615971547e-05,
+      "loss": 0.7961,
+      "step": 1996
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4324275309651531e-05,
+      "loss": 0.7888,
+      "step": 1997
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4318762387277646e-05,
+      "loss": 0.7842,
+      "step": 1998
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4313247850910168e-05,
+      "loss": 0.7917,
+      "step": 1999
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430773170260997e-05,
+      "loss": 0.7617,
+      "step": 2000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4302213944438525e-05,
+      "loss": 0.7937,
+      "step": 2001
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4296694578457917e-05,
+      "loss": 0.8162,
+      "step": 2002
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429117360673082e-05,
+      "loss": 0.7725,
+      "step": 2003
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4285651031320517e-05,
+      "loss": 0.8132,
+      "step": 2004
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4280126854290888e-05,
+      "loss": 0.7993,
+      "step": 2005
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4274601077706404e-05,
+      "loss": 0.7939,
+      "step": 2006
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4269073703632148e-05,
+      "loss": 0.7605,
+      "step": 2007
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4263544734133788e-05,
+      "loss": 0.7908,
+      "step": 2008
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4258014171277588e-05,
+      "loss": 0.7971,
+      "step": 2009
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4252482017130415e-05,
+      "loss": 0.7854,
+      "step": 2010
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4246948273759731e-05,
+      "loss": 0.7944,
+      "step": 2011
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4241412943233583e-05,
+      "loss": 0.7604,
+      "step": 2012
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4235876027620621e-05,
+      "loss": 0.7812,
+      "step": 2013
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423033752899008e-05,
+      "loss": 0.7842,
+      "step": 2014
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4224797449411787e-05,
+      "loss": 0.8264,
+      "step": 2015
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4219255790956168e-05,
+      "loss": 0.7749,
+      "step": 2016
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4213712555694227e-05,
+      "loss": 0.7976,
+      "step": 2017
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.420816774569757e-05,
+      "loss": 0.8115,
+      "step": 2018
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4202621363038383e-05,
+      "loss": 0.7596,
+      "step": 2019
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4197073409789435e-05,
+      "loss": 0.8291,
+      "step": 2020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.41915238880241e-05,
+      "loss": 0.8167,
+      "step": 2021
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4185972799816316e-05,
+      "loss": 0.7656,
+      "step": 2022
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4180420147240627e-05,
+      "loss": 0.8162,
+      "step": 2023
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4174865932372148e-05,
+      "loss": 0.8022,
+      "step": 2024
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4169310157286577e-05,
+      "loss": 0.7888,
+      "step": 2025
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4163752824060206e-05,
+      "loss": 0.7769,
+      "step": 2026
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4158193934769904e-05,
+      "loss": 0.813,
+      "step": 2027
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4152633491493115e-05,
+      "loss": 0.7859,
+      "step": 2028
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4147071496307876e-05,
+      "loss": 0.7993,
+      "step": 2029
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.414150795129279e-05,
+      "loss": 0.7839,
+      "step": 2030
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4135942858527053e-05,
+      "loss": 0.7656,
+      "step": 2031
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4130376220090431e-05,
+      "loss": 0.7693,
+      "step": 2032
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4124808038063266e-05,
+      "loss": 0.8093,
+      "step": 2033
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4119238314526489e-05,
+      "loss": 0.8162,
+      "step": 2034
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4113667051561594e-05,
+      "loss": 0.8335,
+      "step": 2035
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410809425125065e-05,
+      "loss": 0.7847,
+      "step": 2036
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4102519915676314e-05,
+      "loss": 0.7744,
+      "step": 2037
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4096944046921803e-05,
+      "loss": 0.771,
+      "step": 2038
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4091366647070914e-05,
+      "loss": 0.7908,
+      "step": 2039
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4085787718208016e-05,
+      "loss": 0.7681,
+      "step": 2040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4080207262418043e-05,
+      "loss": 0.8003,
+      "step": 2041
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4074625281786507e-05,
+      "loss": 0.8267,
+      "step": 2042
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406904177839949e-05,
+      "loss": 0.7727,
+      "step": 2043
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4063456754343636e-05,
+      "loss": 0.7839,
+      "step": 2044
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4057870211706166e-05,
+      "loss": 0.7686,
+      "step": 2045
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405228215257486e-05,
+      "loss": 0.7791,
+      "step": 2046
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4046692579038074e-05,
+      "loss": 0.7483,
+      "step": 2047
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.404110149318472e-05,
+      "loss": 0.8098,
+      "step": 2048
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4035508897104282e-05,
+      "loss": 0.7878,
+      "step": 2049
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4029914792886807e-05,
+      "loss": 0.7795,
+      "step": 2050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4024319182622905e-05,
+      "loss": 0.8267,
+      "step": 2051
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.401872206840375e-05,
+      "loss": 0.825,
+      "step": 2052
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4013123452321073e-05,
+      "loss": 0.7944,
+      "step": 2053
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4007523336467174e-05,
+      "loss": 0.8086,
+      "step": 2054
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4001921722934909e-05,
+      "loss": 0.7976,
+      "step": 2055
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3996318613817697e-05,
+      "loss": 0.7322,
+      "step": 2056
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3990714011209505e-05,
+      "loss": 0.8281,
+      "step": 2057
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3985107917204879e-05,
+      "loss": 0.7534,
+      "step": 2058
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.39795003338989e-05,
+      "loss": 0.783,
+      "step": 2059
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3973891263387218e-05,
+      "loss": 0.8188,
+      "step": 2060
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3968280707766038e-05,
+      "loss": 0.7595,
+      "step": 2061
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3962668669132115e-05,
+      "loss": 0.7563,
+      "step": 2062
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3957055149582763e-05,
+      "loss": 0.769,
+      "step": 2063
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3951440151215853e-05,
+      "loss": 0.7644,
+      "step": 2064
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3945823676129797e-05,
+      "loss": 0.8164,
+      "step": 2065
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3940205726423568e-05,
+      "loss": 0.8105,
+      "step": 2066
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393458630419669e-05,
+      "loss": 0.7983,
+      "step": 2067
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3928965411549227e-05,
+      "loss": 0.7964,
+      "step": 2068
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.392334305058181e-05,
+      "loss": 0.8005,
+      "step": 2069
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3917719223395603e-05,
+      "loss": 0.7484,
+      "step": 2070
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912093932092326e-05,
+      "loss": 0.8162,
+      "step": 2071
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3906467178774243e-05,
+      "loss": 0.8086,
+      "step": 2072
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3900838965544168e-05,
+      "loss": 0.7579,
+      "step": 2073
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3895209294505458e-05,
+      "loss": 0.8123,
+      "step": 2074
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3889578167762014e-05,
+      "loss": 0.7832,
+      "step": 2075
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3883945587418281e-05,
+      "loss": 0.8174,
+      "step": 2076
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387831155557925e-05,
+      "loss": 0.8105,
+      "step": 2077
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3872676074350451e-05,
+      "loss": 0.7585,
+      "step": 2078
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.386703914583796e-05,
+      "loss": 0.7876,
+      "step": 2079
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.386140077214839e-05,
+      "loss": 0.7656,
+      "step": 2080
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3855760955388894e-05,
+      "loss": 0.7654,
+      "step": 2081
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3850119697667168e-05,
+      "loss": 0.8145,
+      "step": 2082
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3844477001091446e-05,
+      "loss": 0.7622,
+      "step": 2083
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383883286777049e-05,
+      "loss": 0.8127,
+      "step": 2084
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3833187299813613e-05,
+      "loss": 0.7913,
+      "step": 2085
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3827540299330658e-05,
+      "loss": 0.8137,
+      "step": 2086
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3821891868432e-05,
+      "loss": 0.7761,
+      "step": 2087
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3816242009228558e-05,
+      "loss": 0.8047,
+      "step": 2088
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3810590723831772e-05,
+      "loss": 0.7832,
+      "step": 2089
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3804938014353627e-05,
+      "loss": 0.7778,
+      "step": 2090
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3799283882906626e-05,
+      "loss": 0.7837,
+      "step": 2091
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3793628331603823e-05,
+      "loss": 0.7913,
+      "step": 2092
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3787971362558787e-05,
+      "loss": 0.7849,
+      "step": 2093
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3782312977885624e-05,
+      "loss": 0.833,
+      "step": 2094
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3776653179698964e-05,
+      "loss": 0.8132,
+      "step": 2095
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.377099197011397e-05,
+      "loss": 0.762,
+      "step": 2096
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3765329351246328e-05,
+      "loss": 0.7578,
+      "step": 2097
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3759665325212256e-05,
+      "loss": 0.7922,
+      "step": 2098
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3753999894128494e-05,
+      "loss": 0.7957,
+      "step": 2099
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3748333060112306e-05,
+      "loss": 0.7744,
+      "step": 2100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3742664825281488e-05,
+      "loss": 0.8037,
+      "step": 2101
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3736995191754348e-05,
+      "loss": 0.7576,
+      "step": 2102
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3731324161649728e-05,
+      "loss": 0.8003,
+      "step": 2103
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3725651737086982e-05,
+      "loss": 0.7759,
+      "step": 2104
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3719977920185992e-05,
+      "loss": 0.7805,
+      "step": 2105
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3714302713067159e-05,
+      "loss": 0.7598,
+      "step": 2106
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3708626117851398e-05,
+      "loss": 0.8213,
+      "step": 2107
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.370294813666015e-05,
+      "loss": 0.8384,
+      "step": 2108
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3697268771615378e-05,
+      "loss": 0.8188,
+      "step": 2109
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3691588024839544e-05,
+      "loss": 0.7576,
+      "step": 2110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3685905898455648e-05,
+      "loss": 0.8245,
+      "step": 2111
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3680222394587188e-05,
+      "loss": 0.7988,
+      "step": 2112
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3674537515358188e-05,
+      "loss": 0.8101,
+      "step": 2113
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3668851262893182e-05,
+      "loss": 0.7815,
+      "step": 2114
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3663163639317216e-05,
+      "loss": 0.8003,
+      "step": 2115
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365747464675585e-05,
+      "loss": 0.8159,
+      "step": 2116
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365178428733516e-05,
+      "loss": 0.791,
+      "step": 2117
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3646092563181722e-05,
+      "loss": 0.8137,
+      "step": 2118
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364039947642263e-05,
+      "loss": 0.7559,
+      "step": 2119
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3634705029185488e-05,
+      "loss": 0.8145,
+      "step": 2120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3629009223598405e-05,
+      "loss": 0.8037,
+      "step": 2121
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3623312061789995e-05,
+      "loss": 0.7468,
+      "step": 2122
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361761354588939e-05,
+      "loss": 0.7656,
+      "step": 2123
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3611913678026215e-05,
+      "loss": 0.7996,
+      "step": 2124
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3606212460330606e-05,
+      "loss": 0.7583,
+      "step": 2125
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3600509894933203e-05,
+      "loss": 0.772,
+      "step": 2126
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3594805983965152e-05,
+      "loss": 0.8049,
+      "step": 2127
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.35891007295581e-05,
+      "loss": 0.7871,
+      "step": 2128
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3583394133844192e-05,
+      "loss": 0.8076,
+      "step": 2129
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3577686198956081e-05,
+      "loss": 0.8005,
+      "step": 2130
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.357197692702692e-05,
+      "loss": 0.7712,
+      "step": 2131
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3566266320190353e-05,
+      "loss": 0.7866,
+      "step": 2132
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3560554380580536e-05,
+      "loss": 0.7844,
+      "step": 2133
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.355484111033211e-05,
+      "loss": 0.8477,
+      "step": 2134
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3549126511580222e-05,
+      "loss": 0.8228,
+      "step": 2135
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3543410586460512e-05,
+      "loss": 0.7444,
+      "step": 2136
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3537693337109116e-05,
+      "loss": 0.7754,
+      "step": 2137
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3531974765662672e-05,
+      "loss": 0.7954,
+      "step": 2138
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3526254874258303e-05,
+      "loss": 0.8159,
+      "step": 2139
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352053366503362e-05,
+      "loss": 0.7957,
+      "step": 2140
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3514811140126742e-05,
+      "loss": 0.7751,
+      "step": 2141
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.350908730167627e-05,
+      "loss": 0.8081,
+      "step": 2142
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3503362151821299e-05,
+      "loss": 0.8096,
+      "step": 2143
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349763569270141e-05,
+      "loss": 0.7876,
+      "step": 2144
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349190792645668e-05,
+      "loss": 0.8245,
+      "step": 2145
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.348617885522767e-05,
+      "loss": 0.762,
+      "step": 2146
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3480448481155431e-05,
+      "loss": 0.7961,
+      "step": 2147
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3474716806381492e-05,
+      "loss": 0.8257,
+      "step": 2148
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3468983833047884e-05,
+      "loss": 0.8311,
+      "step": 2149
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.346324956329711e-05,
+      "loss": 0.7771,
+      "step": 2150
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3457513999272168e-05,
+      "loss": 0.7803,
+      "step": 2151
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3451777143116527e-05,
+      "loss": 0.7759,
+      "step": 2152
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3446038996974149e-05,
+      "loss": 0.7595,
+      "step": 2153
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3440299562989476e-05,
+      "loss": 0.7622,
+      "step": 2154
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343455884330743e-05,
+      "loss": 0.8179,
+      "step": 2155
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3428816840073413e-05,
+      "loss": 0.7856,
+      "step": 2156
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3423073555433308e-05,
+      "loss": 0.7808,
+      "step": 2157
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417328991533474e-05,
+      "loss": 0.7798,
+      "step": 2158
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.341158315052075e-05,
+      "loss": 0.7861,
+      "step": 2159
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3405836034542459e-05,
+      "loss": 0.7817,
+      "step": 2160
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3400087645746389e-05,
+      "loss": 0.738,
+      "step": 2161
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3394337986280813e-05,
+      "loss": 0.7861,
+      "step": 2162
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3388587058294466e-05,
+      "loss": 0.7711,
+      "step": 2163
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3382834863936575e-05,
+      "loss": 0.7732,
+      "step": 2164
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3377081405356828e-05,
+      "loss": 0.7842,
+      "step": 2165
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3371326684705386e-05,
+      "loss": 0.7749,
+      "step": 2166
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3365570704132886e-05,
+      "loss": 0.7856,
+      "step": 2167
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3359813465790437e-05,
+      "loss": 0.822,
+      "step": 2168
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.335405497182961e-05,
+      "loss": 0.8007,
+      "step": 2169
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3348295224402455e-05,
+      "loss": 0.7961,
+      "step": 2170
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342534225661482e-05,
+      "loss": 0.7764,
+      "step": 2171
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3336771977759675e-05,
+      "loss": 0.7864,
+      "step": 2172
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3331008482850481e-05,
+      "loss": 0.8083,
+      "step": 2173
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3325243743087812e-05,
+      "loss": 0.7939,
+      "step": 2174
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.331947776062605e-05,
+      "loss": 0.7986,
+      "step": 2175
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.331371053762004e-05,
+      "loss": 0.8047,
+      "step": 2176
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3307942076225086e-05,
+      "loss": 0.7988,
+      "step": 2177
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3302172378596964e-05,
+      "loss": 0.8293,
+      "step": 2178
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3296401446891903e-05,
+      "loss": 0.7903,
+      "step": 2179
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3290629283266592e-05,
+      "loss": 0.7703,
+      "step": 2180
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3284855889878192e-05,
+      "loss": 0.7488,
+      "step": 2181
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3279081268884313e-05,
+      "loss": 0.7576,
+      "step": 2182
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3273305422443028e-05,
+      "loss": 0.7708,
+      "step": 2183
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3267528352712869e-05,
+      "loss": 0.7739,
+      "step": 2184
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3261750061852818e-05,
+      "loss": 0.8213,
+      "step": 2185
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3255970552022327e-05,
+      "loss": 0.8057,
+      "step": 2186
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250189825381288e-05,
+      "loss": 0.8462,
+      "step": 2187
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3244407884090058e-05,
+      "loss": 0.7947,
+      "step": 2188
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3238624730309442e-05,
+      "loss": 0.7859,
+      "step": 2189
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3232840366200706e-05,
+      "loss": 0.7864,
+      "step": 2190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322705479392556e-05,
+      "loss": 0.7998,
+      "step": 2191
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3221268015646173e-05,
+      "loss": 0.8179,
+      "step": 2192
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3215480033525157e-05,
+      "loss": 0.7883,
+      "step": 2193
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3209690849725578e-05,
+      "loss": 0.7539,
+      "step": 2194
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3203900466410951e-05,
+      "loss": 0.7725,
+      "step": 2195
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3198108885745236e-05,
+      "loss": 0.752,
+      "step": 2196
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3192316109892846e-05,
+      "loss": 0.7493,
+      "step": 2197
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3186522141018641e-05,
+      "loss": 0.8108,
+      "step": 2198
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3180726981287922e-05,
+      "loss": 0.8179,
+      "step": 2199
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3174930632866433e-05,
+      "loss": 0.823,
+      "step": 2200
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316913309792037e-05,
+      "loss": 0.7761,
+      "step": 2201
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316333437861637e-05,
+      "loss": 0.8008,
+      "step": 2202
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3157534477121505e-05,
+      "loss": 0.7871,
+      "step": 2203
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31517333956033e-05,
+      "loss": 0.7634,
+      "step": 2204
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3145931136229719e-05,
+      "loss": 0.7839,
+      "step": 2205
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3140127701169154e-05,
+      "loss": 0.7644,
+      "step": 2206
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3134323092590452e-05,
+      "loss": 0.7703,
+      "step": 2207
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3128517312662895e-05,
+      "loss": 0.7942,
+      "step": 2208
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312271036355619e-05,
+      "loss": 0.7651,
+      "step": 2209
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3116902247440501e-05,
+      "loss": 0.7336,
+      "step": 2210
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311109296648641e-05,
+      "loss": 0.7407,
+      "step": 2211
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3105282522864946e-05,
+      "loss": 0.7844,
+      "step": 2212
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309947091874757e-05,
+      "loss": 0.7949,
+      "step": 2213
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3093658156306173e-05,
+      "loss": 0.7556,
+      "step": 2214
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3087844237713084e-05,
+      "loss": 0.7827,
+      "step": 2215
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3082029165141059e-05,
+      "loss": 0.8101,
+      "step": 2216
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3076212940763287e-05,
+      "loss": 0.8215,
+      "step": 2217
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3070395566753391e-05,
+      "loss": 0.7834,
+      "step": 2218
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3064577045285418e-05,
+      "loss": 0.7495,
+      "step": 2219
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3058757378533845e-05,
+      "loss": 0.7888,
+      "step": 2220
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3052936568673586e-05,
+      "loss": 0.7915,
+      "step": 2221
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.304711461787997e-05,
+      "loss": 0.7905,
+      "step": 2222
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.304129152832876e-05,
+      "loss": 0.7822,
+      "step": 2223
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3035467302196135e-05,
+      "loss": 0.8096,
+      "step": 2224
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3029641941658708e-05,
+      "loss": 0.7754,
+      "step": 2225
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302381544889352e-05,
+      "loss": 0.8293,
+      "step": 2226
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3017987826078022e-05,
+      "loss": 0.7937,
+      "step": 2227
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3012159075390095e-05,
+      "loss": 0.7764,
+      "step": 2228
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3006329199008043e-05,
+      "loss": 0.7886,
+      "step": 2229
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3000498199110585e-05,
+      "loss": 0.8247,
+      "step": 2230
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2994666077876865e-05,
+      "loss": 0.7874,
+      "step": 2231
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2988832837486445e-05,
+      "loss": 0.7903,
+      "step": 2232
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.29829984801193e-05,
+      "loss": 0.7307,
+      "step": 2233
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2977163007955829e-05,
+      "loss": 0.7971,
+      "step": 2234
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2971326423176849e-05,
+      "loss": 0.8086,
+      "step": 2235
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2965488727963581e-05,
+      "loss": 0.7764,
+      "step": 2236
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2959649924497679e-05,
+      "loss": 0.7947,
+      "step": 2237
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2953810014961194e-05,
+      "loss": 0.7561,
+      "step": 2238
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2947969001536601e-05,
+      "loss": 0.7981,
+      "step": 2239
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2942126886406785e-05,
+      "loss": 0.8403,
+      "step": 2240
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2936283671755039e-05,
+      "loss": 0.7778,
+      "step": 2241
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2930439359765068e-05,
+      "loss": 0.8025,
+      "step": 2242
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2924593952620997e-05,
+      "loss": 0.76,
+      "step": 2243
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2918747452507343e-05,
+      "loss": 0.8091,
+      "step": 2244
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.291289986160905e-05,
+      "loss": 0.8115,
+      "step": 2245
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.290705118211145e-05,
+      "loss": 0.7778,
+      "step": 2246
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29012014162003e-05,
+      "loss": 0.8047,
+      "step": 2247
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2895350566061751e-05,
+      "loss": 0.7664,
+      "step": 2248
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2889498633882367e-05,
+      "loss": 0.8218,
+      "step": 2249
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288364562184911e-05,
+      "loss": 0.7576,
+      "step": 2250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2877791532149351e-05,
+      "loss": 0.79,
+      "step": 2251
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2871936366970856e-05,
+      "loss": 0.7542,
+      "step": 2252
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2866080128501804e-05,
+      "loss": 0.7976,
+      "step": 2253
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.286022281893076e-05,
+      "loss": 0.8047,
+      "step": 2254
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2854364440446712e-05,
+      "loss": 0.7695,
+      "step": 2255
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2848504995239022e-05,
+      "loss": 0.7837,
+      "step": 2256
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2842644485497466e-05,
+      "loss": 0.7603,
+      "step": 2257
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2836782913412219e-05,
+      "loss": 0.7612,
+      "step": 2258
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2830920281173845e-05,
+      "loss": 0.75,
+      "step": 2259
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2825056590973302e-05,
+      "loss": 0.7712,
+      "step": 2260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2819191845001957e-05,
+      "loss": 0.7676,
+      "step": 2261
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2813326045451557e-05,
+      "loss": 0.8181,
+      "step": 2262
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2807459194514257e-05,
+      "loss": 0.8162,
+      "step": 2263
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2801591294382589e-05,
+      "loss": 0.7793,
+      "step": 2264
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279572234724949e-05,
+      "loss": 0.7856,
+      "step": 2265
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2789852355308283e-05,
+      "loss": 0.7751,
+      "step": 2266
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2783981320752683e-05,
+      "loss": 0.8474,
+      "step": 2267
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2778109245776791e-05,
+      "loss": 0.804,
+      "step": 2268
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.27722361325751e-05,
+      "loss": 0.7671,
+      "step": 2269
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276636198334249e-05,
+      "loss": 0.7664,
+      "step": 2270
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2760486800274227e-05,
+      "loss": 0.8127,
+      "step": 2271
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2754610585565974e-05,
+      "loss": 0.7908,
+      "step": 2272
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2748733341413759e-05,
+      "loss": 0.7673,
+      "step": 2273
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2742855070014015e-05,
+      "loss": 0.7944,
+      "step": 2274
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2736975773563544e-05,
+      "loss": 0.8007,
+      "step": 2275
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2731095454259542e-05,
+      "loss": 0.7544,
+      "step": 2276
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2725214114299576e-05,
+      "loss": 0.7349,
+      "step": 2277
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2719331755881608e-05,
+      "loss": 0.7476,
+      "step": 2278
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2713448381203968e-05,
+      "loss": 0.7419,
+      "step": 2279
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2707563992465378e-05,
+      "loss": 0.7258,
+      "step": 2280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2701678591864921e-05,
+      "loss": 0.8315,
+      "step": 2281
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2695792181602085e-05,
+      "loss": 0.7554,
+      "step": 2282
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2689904763876709e-05,
+      "loss": 0.8049,
+      "step": 2283
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2684016340889017e-05,
+      "loss": 0.804,
+      "step": 2284
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2678126914839621e-05,
+      "loss": 0.8059,
+      "step": 2285
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2672236487929488e-05,
+      "loss": 0.7773,
+      "step": 2286
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2666345062359974e-05,
+      "loss": 0.7556,
+      "step": 2287
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2660452640332807e-05,
+      "loss": 0.8015,
+      "step": 2288
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2654559224050074e-05,
+      "loss": 0.7808,
+      "step": 2289
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2648664815714251e-05,
+      "loss": 0.7839,
+      "step": 2290
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2642769417528173e-05,
+      "loss": 0.8154,
+      "step": 2291
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263687303169505e-05,
+      "loss": 0.8013,
+      "step": 2292
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2630975660418456e-05,
+      "loss": 0.8223,
+      "step": 2293
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.262507730590234e-05,
+      "loss": 0.8298,
+      "step": 2294
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.261917797035102e-05,
+      "loss": 0.7729,
+      "step": 2295
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.261327765596917e-05,
+      "loss": 0.7815,
+      "step": 2296
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2607376364961838e-05,
+      "loss": 0.7671,
+      "step": 2297
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2601474099534437e-05,
+      "loss": 0.7266,
+      "step": 2298
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2595570861892739e-05,
+      "loss": 0.8203,
+      "step": 2299
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2589666654242883e-05,
+      "loss": 0.802,
+      "step": 2300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2583761478791367e-05,
+      "loss": 0.7712,
+      "step": 2301
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2577855337745059e-05,
+      "loss": 0.8005,
+      "step": 2302
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2571948233311184e-05,
+      "loss": 0.8132,
+      "step": 2303
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2566040167697322e-05,
+      "loss": 0.792,
+      "step": 2304
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2560131143111412e-05,
+      "loss": 0.7327,
+      "step": 2305
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2554221161761761e-05,
+      "loss": 0.7668,
+      "step": 2306
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2548310225857026e-05,
+      "loss": 0.7612,
+      "step": 2307
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2542398337606221e-05,
+      "loss": 0.7932,
+      "step": 2308
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2536485499218718e-05,
+      "loss": 0.7898,
+      "step": 2309
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2530571712904246e-05,
+      "loss": 0.7668,
+      "step": 2310
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2524656980872883e-05,
+      "loss": 0.7991,
+      "step": 2311
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2518741305335062e-05,
+      "loss": 0.7717,
+      "step": 2312
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2512824688501572e-05,
+      "loss": 0.8259,
+      "step": 2313
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2506907132583553e-05,
+      "loss": 0.7964,
+      "step": 2314
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2500988639792486e-05,
+      "loss": 0.7656,
+      "step": 2315
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.249506921234022e-05,
+      "loss": 0.8022,
+      "step": 2316
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2489148852438944e-05,
+      "loss": 0.7698,
+      "step": 2317
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2483227562301187e-05,
+      "loss": 0.7781,
+      "step": 2318
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2477305344139842e-05,
+      "loss": 0.7947,
+      "step": 2319
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2471382200168139e-05,
+      "loss": 0.7666,
+      "step": 2320
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246545813259965e-05,
+      "loss": 0.8352,
+      "step": 2321
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2459533143648306e-05,
+      "loss": 0.7849,
+      "step": 2322
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2453607235528366e-05,
+      "loss": 0.7935,
+      "step": 2323
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2447680410454448e-05,
+      "loss": 0.8052,
+      "step": 2324
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2441752670641504e-05,
+      "loss": 0.7604,
+      "step": 2325
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2435824018304827e-05,
+      "loss": 0.7869,
+      "step": 2326
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2429894455660057e-05,
+      "loss": 0.7856,
+      "step": 2327
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2423963984923166e-05,
+      "loss": 0.7964,
+      "step": 2328
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2418032608310472e-05,
+      "loss": 0.8032,
+      "step": 2329
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2412100328038626e-05,
+      "loss": 0.7766,
+      "step": 2330
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2406167146324628e-05,
+      "loss": 0.7959,
+      "step": 2331
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2400233065385801e-05,
+      "loss": 0.7434,
+      "step": 2332
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2394298087439813e-05,
+      "loss": 0.7817,
+      "step": 2333
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238836221470466e-05,
+      "loss": 0.7344,
+      "step": 2334
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2382425449398685e-05,
+      "loss": 0.7637,
+      "step": 2335
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2376487793740548e-05,
+      "loss": 0.7637,
+      "step": 2336
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2370549249949254e-05,
+      "loss": 0.7761,
+      "step": 2337
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236460982024413e-05,
+      "loss": 0.771,
+      "step": 2338
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2358669506844852e-05,
+      "loss": 0.8296,
+      "step": 2339
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2352728311971402e-05,
+      "loss": 0.7974,
+      "step": 2340
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2346786237844112e-05,
+      "loss": 0.7563,
+      "step": 2341
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2340843286683632e-05,
+      "loss": 0.7837,
+      "step": 2342
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.233489946071094e-05,
+      "loss": 0.8054,
+      "step": 2343
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2328954762147347e-05,
+      "loss": 0.7292,
+      "step": 2344
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2323009193214478e-05,
+      "loss": 0.8245,
+      "step": 2345
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2317062756134302e-05,
+      "loss": 0.7634,
+      "step": 2346
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2311115453129095e-05,
+      "loss": 0.7786,
+      "step": 2347
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2305167286421462e-05,
+      "loss": 0.7832,
+      "step": 2348
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.229921825823434e-05,
+      "loss": 0.8503,
+      "step": 2349
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2293268370790972e-05,
+      "loss": 0.7754,
+      "step": 2350
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2287317626314935e-05,
+      "loss": 0.8083,
+      "step": 2351
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.228136602703012e-05,
+      "loss": 0.8,
+      "step": 2352
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2275413575160738e-05,
+      "loss": 0.8066,
+      "step": 2353
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2269460272931321e-05,
+      "loss": 0.8162,
+      "step": 2354
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.226350612256672e-05,
+      "loss": 0.8098,
+      "step": 2355
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2257551126292093e-05,
+      "loss": 0.8096,
+      "step": 2356
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225159528633293e-05,
+      "loss": 0.7673,
+      "step": 2357
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2245638604915023e-05,
+      "loss": 0.7551,
+      "step": 2358
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2239681084264483e-05,
+      "loss": 0.8218,
+      "step": 2359
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2233722726607735e-05,
+      "loss": 0.7407,
+      "step": 2360
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.222776353417152e-05,
+      "loss": 0.752,
+      "step": 2361
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2221803509182886e-05,
+      "loss": 0.7664,
+      "step": 2362
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2215842653869192e-05,
+      "loss": 0.7898,
+      "step": 2363
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.220988097045811e-05,
+      "loss": 0.811,
+      "step": 2364
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2203918461177621e-05,
+      "loss": 0.8254,
+      "step": 2365
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2197955128256015e-05,
+      "loss": 0.7698,
+      "step": 2366
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.219199097392189e-05,
+      "loss": 0.8064,
+      "step": 2367
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2186026000404146e-05,
+      "loss": 0.7815,
+      "step": 2368
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2180060209931994e-05,
+      "loss": 0.7925,
+      "step": 2369
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2174093604734953e-05,
+      "loss": 0.7864,
+      "step": 2370
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2168126187042843e-05,
+      "loss": 0.7559,
+      "step": 2371
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2162157959085783e-05,
+      "loss": 0.8184,
+      "step": 2372
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2156188923094203e-05,
+      "loss": 0.7822,
+      "step": 2373
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2150219081298828e-05,
+      "loss": 0.8198,
+      "step": 2374
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2144248435930693e-05,
+      "loss": 0.8035,
+      "step": 2375
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2138276989221125e-05,
+      "loss": 0.7878,
+      "step": 2376
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2132304743401751e-05,
+      "loss": 0.7794,
+      "step": 2377
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2126331700704502e-05,
+      "loss": 0.7644,
+      "step": 2378
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2120357863361602e-05,
+      "loss": 0.8379,
+      "step": 2379
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2114383233605575e-05,
+      "loss": 0.8037,
+      "step": 2380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2108407813669241e-05,
+      "loss": 0.811,
+      "step": 2381
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210243160578571e-05,
+      "loss": 0.7817,
+      "step": 2382
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2096454612188393e-05,
+      "loss": 0.7871,
+      "step": 2383
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2090476835110996e-05,
+      "loss": 0.7979,
+      "step": 2384
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2084498276787506e-05,
+      "loss": 0.7944,
+      "step": 2385
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2078518939452218e-05,
+      "loss": 0.782,
+      "step": 2386
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2072538825339707e-05,
+      "loss": 0.783,
+      "step": 2387
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206655793668484e-05,
+      "loss": 0.7209,
+      "step": 2388
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2060576275722776e-05,
+      "loss": 0.7866,
+      "step": 2389
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205459384468896e-05,
+      "loss": 0.8147,
+      "step": 2390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204861064581913e-05,
+      "loss": 0.8179,
+      "step": 2391
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2042626681349305e-05,
+      "loss": 0.791,
+      "step": 2392
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2036641953515793e-05,
+      "loss": 0.7415,
+      "step": 2393
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2030656464555186e-05,
+      "loss": 0.7693,
+      "step": 2394
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2024670216704364e-05,
+      "loss": 0.7961,
+      "step": 2395
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2018683212200482e-05,
+      "loss": 0.7698,
+      "step": 2396
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2012695453280987e-05,
+      "loss": 0.7515,
+      "step": 2397
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2006706942183605e-05,
+      "loss": 0.7827,
+      "step": 2398
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2000717681146344e-05,
+      "loss": 0.771,
+      "step": 2399
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1994727672407487e-05,
+      "loss": 0.7883,
+      "step": 2400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1988736918205602e-05,
+      "loss": 0.7939,
+      "step": 2401
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1982745420779533e-05,
+      "loss": 0.7761,
+      "step": 2402
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1976753182368406e-05,
+      "loss": 0.7544,
+      "step": 2403
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1970760205211615e-05,
+      "loss": 0.7654,
+      "step": 2404
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1964766491548838e-05,
+      "loss": 0.8169,
+      "step": 2405
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1958772043620028e-05,
+      "loss": 0.8284,
+      "step": 2406
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.195277686366541e-05,
+      "loss": 0.7964,
+      "step": 2407
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.194678095392548e-05,
+      "loss": 0.8093,
+      "step": 2408
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.194078431664101e-05,
+      "loss": 0.8176,
+      "step": 2409
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1934786954053048e-05,
+      "loss": 0.7452,
+      "step": 2410
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1928788868402903e-05,
+      "loss": 0.7483,
+      "step": 2411
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1922790061932159e-05,
+      "loss": 0.7844,
+      "step": 2412
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1916790536882674e-05,
+      "loss": 0.7693,
+      "step": 2413
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191079029549657e-05,
+      "loss": 0.7102,
+      "step": 2414
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1904789340016238e-05,
+      "loss": 0.7808,
+      "step": 2415
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1898787672684335e-05,
+      "loss": 0.8225,
+      "step": 2416
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1892785295743781e-05,
+      "loss": 0.7871,
+      "step": 2417
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1886782211437765e-05,
+      "loss": 0.8253,
+      "step": 2418
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1880778422009744e-05,
+      "loss": 0.8049,
+      "step": 2419
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1874773929703432e-05,
+      "loss": 0.7925,
+      "step": 2420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1868768736762808e-05,
+      "loss": 0.8042,
+      "step": 2421
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1862762845432111e-05,
+      "loss": 0.7988,
+      "step": 2422
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1856756257955848e-05,
+      "loss": 0.7932,
+      "step": 2423
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1850748976578776e-05,
+      "loss": 0.7732,
+      "step": 2424
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1844741003545916e-05,
+      "loss": 0.7959,
+      "step": 2425
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1838732341102555e-05,
+      "loss": 0.7979,
+      "step": 2426
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.183272299149422e-05,
+      "loss": 0.8062,
+      "step": 2427
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.182671295696671e-05,
+      "loss": 0.8149,
+      "step": 2428
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.182070223976608e-05,
+      "loss": 0.7769,
+      "step": 2429
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1814690842138628e-05,
+      "loss": 0.7961,
+      "step": 2430
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1808678766330917e-05,
+      "loss": 0.7957,
+      "step": 2431
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1802666014589761e-05,
+      "loss": 0.7859,
+      "step": 2432
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1796652589162224e-05,
+      "loss": 0.8132,
+      "step": 2433
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1790638492295624e-05,
+      "loss": 0.8049,
+      "step": 2434
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1784623726237528e-05,
+      "loss": 0.8052,
+      "step": 2435
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1778608293235755e-05,
+      "loss": 0.7788,
+      "step": 2436
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1772592195538377e-05,
+      "loss": 0.7903,
+      "step": 2437
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1766575435393704e-05,
+      "loss": 0.8101,
+      "step": 2438
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1760558015050302e-05,
+      "loss": 0.8098,
+      "step": 2439
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1754539936756986e-05,
+      "loss": 0.7957,
+      "step": 2440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1748521202762803e-05,
+      "loss": 0.8257,
+      "step": 2441
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.174250181531706e-05,
+      "loss": 0.7866,
+      "step": 2442
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.8103,
+      "step": 2443
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1730461089069323e-05,
+      "loss": 0.7712,
+      "step": 2444
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1724439754767145e-05,
+      "loss": 0.8108,
+      "step": 2445
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1718417776013044e-05,
+      "loss": 0.8091,
+      "step": 2446
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171239515505754e-05,
+      "loss": 0.821,
+      "step": 2447
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1706371894151382e-05,
+      "loss": 0.7737,
+      "step": 2448
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.170034799554556e-05,
+      "loss": 0.8025,
+      "step": 2449
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.169432346149131e-05,
+      "loss": 0.791,
+      "step": 2450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1688298294240104e-05,
+      "loss": 0.8088,
+      "step": 2451
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168227249604364e-05,
+      "loss": 0.7949,
+      "step": 2452
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1676246069153863e-05,
+      "loss": 0.7969,
+      "step": 2453
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.167021901582295e-05,
+      "loss": 0.8083,
+      "step": 2454
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1664191338303312e-05,
+      "loss": 0.8088,
+      "step": 2455
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1658163038847589e-05,
+      "loss": 0.719,
+      "step": 2456
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165213411970866e-05,
+      "loss": 0.7693,
+      "step": 2457
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1646104583139632e-05,
+      "loss": 0.832,
+      "step": 2458
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1640074431393846e-05,
+      "loss": 0.8284,
+      "step": 2459
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1634043666724868e-05,
+      "loss": 0.8074,
+      "step": 2460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1628012291386495e-05,
+      "loss": 0.8242,
+      "step": 2461
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1621980307632753e-05,
+      "loss": 0.8074,
+      "step": 2462
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1615947717717895e-05,
+      "loss": 0.7893,
+      "step": 2463
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1609914523896399e-05,
+      "loss": 0.8345,
+      "step": 2464
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1603880728422974e-05,
+      "loss": 0.7891,
+      "step": 2465
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1597846333552549e-05,
+      "loss": 0.7947,
+      "step": 2466
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1591811341540278e-05,
+      "loss": 0.7954,
+      "step": 2467
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1585775754641533e-05,
+      "loss": 0.7979,
+      "step": 2468
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1579739575111924e-05,
+      "loss": 0.7668,
+      "step": 2469
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1573702805207261e-05,
+      "loss": 0.792,
+      "step": 2470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1567665447183594e-05,
+      "loss": 0.7898,
+      "step": 2471
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1561627503297177e-05,
+      "loss": 0.8318,
+      "step": 2472
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1555588975804497e-05,
+      "loss": 0.7686,
+      "step": 2473
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1549549866962251e-05,
+      "loss": 0.7869,
+      "step": 2474
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1543510179027355e-05,
+      "loss": 0.8032,
+      "step": 2475
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1537469914256937e-05,
+      "loss": 0.8179,
+      "step": 2476
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1531429074908352e-05,
+      "loss": 0.7935,
+      "step": 2477
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1525387663239153e-05,
+      "loss": 0.7754,
+      "step": 2478
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1519345681507125e-05,
+      "loss": 0.8005,
+      "step": 2479
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1513303131970254e-05,
+      "loss": 0.7649,
+      "step": 2480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1507260016886742e-05,
+      "loss": 0.7957,
+      "step": 2481
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1501216338515003e-05,
+      "loss": 0.7971,
+      "step": 2482
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1495172099113658e-05,
+      "loss": 0.782,
+      "step": 2483
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148912730094154e-05,
+      "loss": 0.7878,
+      "step": 2484
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1483081946257695e-05,
+      "loss": 0.7949,
+      "step": 2485
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.147703603732137e-05,
+      "loss": 0.7625,
+      "step": 2486
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.147098957639202e-05,
+      "loss": 0.7957,
+      "step": 2487
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1464942565729313e-05,
+      "loss": 0.8137,
+      "step": 2488
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1458895007593116e-05,
+      "loss": 0.7551,
+      "step": 2489
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1452846904243501e-05,
+      "loss": 0.8086,
+      "step": 2490
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.144679825794075e-05,
+      "loss": 0.823,
+      "step": 2491
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1440749070945334e-05,
+      "loss": 0.7961,
+      "step": 2492
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1434699345517946e-05,
+      "loss": 0.7632,
+      "step": 2493
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1428649083919459e-05,
+      "loss": 0.7908,
+      "step": 2494
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1422598288410962e-05,
+      "loss": 0.8423,
+      "step": 2495
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.141654696125374e-05,
+      "loss": 0.7642,
+      "step": 2496
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1410495104709271e-05,
+      "loss": 0.7993,
+      "step": 2497
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1404442721039241e-05,
+      "loss": 0.7925,
+      "step": 2498
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1398389812505522e-05,
+      "loss": 0.7646,
+      "step": 2499
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1392336381370184e-05,
+      "loss": 0.7917,
+      "step": 2500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1386282429895501e-05,
+      "loss": 0.7866,
+      "step": 2501
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.138022796034393e-05,
+      "loss": 0.7688,
+      "step": 2502
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1374172974978133e-05,
+      "loss": 0.7888,
+      "step": 2503
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136811747606096e-05,
+      "loss": 0.7576,
+      "step": 2504
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1362061465855445e-05,
+      "loss": 0.7957,
+      "step": 2505
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1356004946624825e-05,
+      "loss": 0.7981,
+      "step": 2506
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1349947920632524e-05,
+      "loss": 0.7827,
+      "step": 2507
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.134389039014215e-05,
+      "loss": 0.8186,
+      "step": 2508
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.13378323574175e-05,
+      "loss": 0.7898,
+      "step": 2509
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1331773824722569e-05,
+      "loss": 0.7576,
+      "step": 2510
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1325714794321532e-05,
+      "loss": 0.821,
+      "step": 2511
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1319655268478746e-05,
+      "loss": 0.8022,
+      "step": 2512
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1313595249458756e-05,
+      "loss": 0.8159,
+      "step": 2513
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1307534739526295e-05,
+      "loss": 0.7998,
+      "step": 2514
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1301473740946278e-05,
+      "loss": 0.8115,
+      "step": 2515
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1295412255983795e-05,
+      "loss": 0.7449,
+      "step": 2516
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.128935028690413e-05,
+      "loss": 0.7822,
+      "step": 2517
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128328783597274e-05,
+      "loss": 0.8013,
+      "step": 2518
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1277224905455266e-05,
+      "loss": 0.7705,
+      "step": 2519
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1271161497617526e-05,
+      "loss": 0.7915,
+      "step": 2520
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1265097614725513e-05,
+      "loss": 0.772,
+      "step": 2521
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1259033259045405e-05,
+      "loss": 0.7891,
+      "step": 2522
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125296843284355e-05,
+      "loss": 0.7976,
+      "step": 2523
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1246903138386478e-05,
+      "loss": 0.7751,
+      "step": 2524
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.124083737794089e-05,
+      "loss": 0.8174,
+      "step": 2525
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1234771153773658e-05,
+      "loss": 0.7891,
+      "step": 2526
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.122870446815184e-05,
+      "loss": 0.7422,
+      "step": 2527
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1222637323342648e-05,
+      "loss": 0.7278,
+      "step": 2528
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1216569721613482e-05,
+      "loss": 0.7412,
+      "step": 2529
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1210501665231902e-05,
+      "loss": 0.7615,
+      "step": 2530
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1204433156465643e-05,
+      "loss": 0.7959,
+      "step": 2531
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1198364197582608e-05,
+      "loss": 0.7922,
+      "step": 2532
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1192294790850871e-05,
+      "loss": 0.832,
+      "step": 2533
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1186224938538666e-05,
+      "loss": 0.8047,
+      "step": 2534
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1180154642914403e-05,
+      "loss": 0.7971,
+      "step": 2535
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.117408390624665e-05,
+      "loss": 0.792,
+      "step": 2536
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1168012730804142e-05,
+      "loss": 0.7627,
+      "step": 2537
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.116194111885578e-05,
+      "loss": 0.7722,
+      "step": 2538
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1155869072670624e-05,
+      "loss": 0.7825,
+      "step": 2539
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1149796594517901e-05,
+      "loss": 0.7727,
+      "step": 2540
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1143723686667e-05,
+      "loss": 0.823,
+      "step": 2541
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1137650351387466e-05,
+      "loss": 0.7832,
+      "step": 2542
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1131576590949005e-05,
+      "loss": 0.8047,
+      "step": 2543
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1125502407621485e-05,
+      "loss": 0.8071,
+      "step": 2544
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1119427803674924e-05,
+      "loss": 0.7988,
+      "step": 2545
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1113352781379508e-05,
+      "loss": 0.8035,
+      "step": 2546
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1107277343005575e-05,
+      "loss": 0.7947,
+      "step": 2547
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1101201490823615e-05,
+      "loss": 0.7781,
+      "step": 2548
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1095125227104275e-05,
+      "loss": 0.8147,
+      "step": 2549
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1089048554118357e-05,
+      "loss": 0.8169,
+      "step": 2550
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1082971474136817e-05,
+      "loss": 0.7925,
+      "step": 2551
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1076893989430761e-05,
+      "loss": 0.7837,
+      "step": 2552
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1070816102271442e-05,
+      "loss": 0.7903,
+      "step": 2553
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.106473781493027e-05,
+      "loss": 0.8002,
+      "step": 2554
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1058659129678809e-05,
+      "loss": 0.7986,
+      "step": 2555
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1052580048788754e-05,
+      "loss": 0.7815,
+      "step": 2556
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1046500574531969e-05,
+      "loss": 0.79,
+      "step": 2557
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1040420709180446e-05,
+      "loss": 0.7864,
+      "step": 2558
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1034340455006343e-05,
+      "loss": 0.7822,
+      "step": 2559
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1028259814281943e-05,
+      "loss": 0.7864,
+      "step": 2560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1022178789279685e-05,
+      "loss": 0.7842,
+      "step": 2561
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.101609738227215e-05,
+      "loss": 0.7593,
+      "step": 2562
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1010015595532066e-05,
+      "loss": 0.7922,
+      "step": 2563
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.100393343133229e-05,
+      "loss": 0.7712,
+      "step": 2564
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0997850891945836e-05,
+      "loss": 0.8025,
+      "step": 2565
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0991767979645847e-05,
+      "loss": 0.8042,
+      "step": 2566
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0985684696705607e-05,
+      "loss": 0.7788,
+      "step": 2567
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0979601045398542e-05,
+      "loss": 0.7991,
+      "step": 2568
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0973517027998216e-05,
+      "loss": 0.7961,
+      "step": 2569
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0967432646778326e-05,
+      "loss": 0.7937,
+      "step": 2570
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.096134790401271e-05,
+      "loss": 0.7888,
+      "step": 2571
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095526280197533e-05,
+      "loss": 0.8206,
+      "step": 2572
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0949177342940298e-05,
+      "loss": 0.7913,
+      "step": 2573
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0943091529181849e-05,
+      "loss": 0.76,
+      "step": 2574
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.093700536297435e-05,
+      "loss": 0.8018,
+      "step": 2575
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0930918846592309e-05,
+      "loss": 0.7908,
+      "step": 2576
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0924831982310354e-05,
+      "loss": 0.7947,
+      "step": 2577
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.091874477240325e-05,
+      "loss": 0.8181,
+      "step": 2578
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0912657219145893e-05,
+      "loss": 0.8171,
+      "step": 2579
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0906569324813291e-05,
+      "loss": 0.7844,
+      "step": 2580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0900481091680606e-05,
+      "loss": 0.8093,
+      "step": 2581
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.08943925220231e-05,
+      "loss": 0.7349,
+      "step": 2582
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0888303618116185e-05,
+      "loss": 0.8013,
+      "step": 2583
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0882214382235375e-05,
+      "loss": 0.8198,
+      "step": 2584
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0876124816656324e-05,
+      "loss": 0.7986,
+      "step": 2585
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0870034923654808e-05,
+      "loss": 0.8005,
+      "step": 2586
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0863944705506716e-05,
+      "loss": 0.748,
+      "step": 2587
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0857854164488066e-05,
+      "loss": 0.7727,
+      "step": 2588
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0851763302874998e-05,
+      "loss": 0.7566,
+      "step": 2589
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0845672122943763e-05,
+      "loss": 0.8274,
+      "step": 2590
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0839580626970736e-05,
+      "loss": 0.7979,
+      "step": 2591
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.083348881723242e-05,
+      "loss": 0.8096,
+      "step": 2592
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0827396696005417e-05,
+      "loss": 0.8413,
+      "step": 2593
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0821304265566459e-05,
+      "loss": 0.7927,
+      "step": 2594
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0815211528192386e-05,
+      "loss": 0.738,
+      "step": 2595
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0809118486160158e-05,
+      "loss": 0.7756,
+      "step": 2596
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0803025141746845e-05,
+      "loss": 0.822,
+      "step": 2597
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079693149722963e-05,
+      "loss": 0.7441,
+      "step": 2598
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0790837554885813e-05,
+      "loss": 0.7798,
+      "step": 2599
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0784743316992803e-05,
+      "loss": 0.7893,
+      "step": 2600
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0778648785828113e-05,
+      "loss": 0.8271,
+      "step": 2601
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.077255396366938e-05,
+      "loss": 0.7852,
+      "step": 2602
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0766458852794331e-05,
+      "loss": 0.7393,
+      "step": 2603
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0760363455480814e-05,
+      "loss": 0.7637,
+      "step": 2604
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0754267774006782e-05,
+      "loss": 0.783,
+      "step": 2605
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0748171810650292e-05,
+      "loss": 0.8,
+      "step": 2606
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0742075567689504e-05,
+      "loss": 0.8135,
+      "step": 2607
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0735979047402696e-05,
+      "loss": 0.749,
+      "step": 2608
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.072988225206823e-05,
+      "loss": 0.8059,
+      "step": 2609
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0723785183964581e-05,
+      "loss": 0.8096,
+      "step": 2610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.071768784537033e-05,
+      "loss": 0.8101,
+      "step": 2611
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.071159023856415e-05,
+      "loss": 0.8147,
+      "step": 2612
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0705492365824818e-05,
+      "loss": 0.7651,
+      "step": 2613
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0699394229431213e-05,
+      "loss": 0.7656,
+      "step": 2614
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069329583166231e-05,
+      "loss": 0.8071,
+      "step": 2615
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0687197174797185e-05,
+      "loss": 0.7959,
+      "step": 2616
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0681098261115002e-05,
+      "loss": 0.7834,
+      "step": 2617
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0674999092895031e-05,
+      "loss": 0.8308,
+      "step": 2618
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0668899672416632e-05,
+      "loss": 0.7888,
+      "step": 2619
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0662800001959259e-05,
+      "loss": 0.7695,
+      "step": 2620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0656700083802457e-05,
+      "loss": 0.7622,
+      "step": 2621
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0650599920225877e-05,
+      "loss": 0.7781,
+      "step": 2622
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0644499513509247e-05,
+      "loss": 0.8083,
+      "step": 2623
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0638398865932391e-05,
+      "loss": 0.8596,
+      "step": 2624
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.063229797977522e-05,
+      "loss": 0.8003,
+      "step": 2625
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0626196857317739e-05,
+      "loss": 0.7424,
+      "step": 2626
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0620095500840043e-05,
+      "loss": 0.8193,
+      "step": 2627
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0613993912622303e-05,
+      "loss": 0.74,
+      "step": 2628
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0607892094944789e-05,
+      "loss": 0.8064,
+      "step": 2629
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0601790050087855e-05,
+      "loss": 0.8164,
+      "step": 2630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0595687780331932e-05,
+      "loss": 0.8181,
+      "step": 2631
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0589585287957544e-05,
+      "loss": 0.739,
+      "step": 2632
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058348257524529e-05,
+      "loss": 0.8071,
+      "step": 2633
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057737964447586e-05,
+      "loss": 0.8081,
+      "step": 2634
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0571276497930016e-05,
+      "loss": 0.8154,
+      "step": 2635
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.056517313788861e-05,
+      "loss": 0.8013,
+      "step": 2636
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0559069566632569e-05,
+      "loss": 0.7361,
+      "step": 2637
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.05529657864429e-05,
+      "loss": 0.7651,
+      "step": 2638
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0546861799600685e-05,
+      "loss": 0.7576,
+      "step": 2639
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0540757608387086e-05,
+      "loss": 0.8149,
+      "step": 2640
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0534653215083343e-05,
+      "loss": 0.8142,
+      "step": 2641
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.052854862197077e-05,
+      "loss": 0.79,
+      "step": 2642
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.052244383133075e-05,
+      "loss": 0.7644,
+      "step": 2643
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0516338845444754e-05,
+      "loss": 0.8157,
+      "step": 2644
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0510233666594311e-05,
+      "loss": 0.7625,
+      "step": 2645
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0504128297061029e-05,
+      "loss": 0.7759,
+      "step": 2646
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0498022739126588e-05,
+      "loss": 0.8101,
+      "step": 2647
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0491916995072737e-05,
+      "loss": 0.7783,
+      "step": 2648
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0485811067181292e-05,
+      "loss": 0.7888,
+      "step": 2649
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0479704957734145e-05,
+      "loss": 0.7561,
+      "step": 2650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0473598669013243e-05,
+      "loss": 0.8152,
+      "step": 2651
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0467492203300615e-05,
+      "loss": 0.7859,
+      "step": 2652
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0461385562878347e-05,
+      "loss": 0.7925,
+      "step": 2653
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0455278750028592e-05,
+      "loss": 0.7422,
+      "step": 2654
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0449171767033571e-05,
+      "loss": 0.7981,
+      "step": 2655
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0443064616175562e-05,
+      "loss": 0.801,
+      "step": 2656
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0436957299736907e-05,
+      "loss": 0.7837,
+      "step": 2657
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0430849820000017e-05,
+      "loss": 0.7896,
+      "step": 2658
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0424742179247361e-05,
+      "loss": 0.7837,
+      "step": 2659
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0418634379761457e-05,
+      "loss": 0.8357,
+      "step": 2660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.04125264238249e-05,
+      "loss": 0.7705,
+      "step": 2661
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0406418313720333e-05,
+      "loss": 0.7881,
+      "step": 2662
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0400310051730455e-05,
+      "loss": 0.7903,
+      "step": 2663
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0394201640138034e-05,
+      "loss": 0.7935,
+      "step": 2664
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0388093081225874e-05,
+      "loss": 0.8088,
+      "step": 2665
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0381984377276853e-05,
+      "loss": 0.7878,
+      "step": 2666
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0375875530573896e-05,
+      "loss": 0.791,
+      "step": 2667
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0369766543399978e-05,
+      "loss": 0.7925,
+      "step": 2668
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0363657418038128e-05,
+      "loss": 0.7983,
+      "step": 2669
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0357548156771433e-05,
+      "loss": 0.7551,
+      "step": 2670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0351438761883023e-05,
+      "loss": 0.8264,
+      "step": 2671
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0345329235656082e-05,
+      "loss": 0.7791,
+      "step": 2672
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0339219580373841e-05,
+      "loss": 0.7913,
+      "step": 2673
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033310979831958e-05,
+      "loss": 0.7964,
+      "step": 2674
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0326999891776628e-05,
+      "loss": 0.8159,
+      "step": 2675
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0320889863028356e-05,
+      "loss": 0.7957,
+      "step": 2676
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0314779714358188e-05,
+      "loss": 0.8276,
+      "step": 2677
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0308669448049584e-05,
+      "loss": 0.7876,
+      "step": 2678
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0302559066386055e-05,
+      "loss": 0.7708,
+      "step": 2679
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0296448571651154e-05,
+      "loss": 0.8059,
+      "step": 2680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0290337966128471e-05,
+      "loss": 0.7705,
+      "step": 2681
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0284227252101646e-05,
+      "loss": 0.7747,
+      "step": 2682
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0278116431854352e-05,
+      "loss": 0.762,
+      "step": 2683
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0272005507670304e-05,
+      "loss": 0.7695,
+      "step": 2684
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026589448183326e-05,
+      "loss": 0.7283,
+      "step": 2685
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0259783356627007e-05,
+      "loss": 0.7681,
+      "step": 2686
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.025367213433538e-05,
+      "loss": 0.7769,
+      "step": 2687
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0247560817242237e-05,
+      "loss": 0.7791,
+      "step": 2688
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0241449407631488e-05,
+      "loss": 0.7527,
+      "step": 2689
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0235337907787064e-05,
+      "loss": 0.7793,
+      "step": 2690
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0229226319992938e-05,
+      "loss": 0.7883,
+      "step": 2691
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0223114646533105e-05,
+      "loss": 0.8096,
+      "step": 2692
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0217002889691608e-05,
+      "loss": 0.782,
+      "step": 2693
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0210891051752508e-05,
+      "loss": 0.7847,
+      "step": 2694
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.02047791349999e-05,
+      "loss": 0.7937,
+      "step": 2695
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0198667141717913e-05,
+      "loss": 0.8259,
+      "step": 2696
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0192555074190696e-05,
+      "loss": 0.7639,
+      "step": 2697
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0186442934702436e-05,
+      "loss": 0.7971,
+      "step": 2698
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0180330725537342e-05,
+      "loss": 0.7515,
+      "step": 2699
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.017421844897964e-05,
+      "loss": 0.7585,
+      "step": 2700
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01681061073136e-05,
+      "loss": 0.7698,
+      "step": 2701
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0161993702823499e-05,
+      "loss": 0.7793,
+      "step": 2702
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0155881237793646e-05,
+      "loss": 0.7827,
+      "step": 2703
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0149768714508374e-05,
+      "loss": 0.7917,
+      "step": 2704
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0143656135252033e-05,
+      "loss": 0.7432,
+      "step": 2705
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0137543502309e-05,
+      "loss": 0.7649,
+      "step": 2706
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013143081796366e-05,
+      "loss": 0.7837,
+      "step": 2707
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.012531808450043e-05,
+      "loss": 0.7925,
+      "step": 2708
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119205304203741e-05,
+      "loss": 0.7397,
+      "step": 2709
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011309247935804e-05,
+      "loss": 0.8052,
+      "step": 2710
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0106979612247789e-05,
+      "loss": 0.79,
+      "step": 2711
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0100866705157475e-05,
+      "loss": 0.7864,
+      "step": 2712
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0094753760371587e-05,
+      "loss": 0.7656,
+      "step": 2713
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0088640780174639e-05,
+      "loss": 0.8086,
+      "step": 2714
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0082527766851151e-05,
+      "loss": 0.7383,
+      "step": 2715
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0076414722685658e-05,
+      "loss": 0.7441,
+      "step": 2716
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0070301649962711e-05,
+      "loss": 0.8032,
+      "step": 2717
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0064188550966858e-05,
+      "loss": 0.7942,
+      "step": 2718
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0058075427982674e-05,
+      "loss": 0.7959,
+      "step": 2719
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0051962283294735e-05,
+      "loss": 0.7659,
+      "step": 2720
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0045849119187622e-05,
+      "loss": 0.7673,
+      "step": 2721
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003973593794593e-05,
+      "loss": 0.7644,
+      "step": 2722
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0033622741854254e-05,
+      "loss": 0.8252,
+      "step": 2723
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0027509533197198e-05,
+      "loss": 0.7878,
+      "step": 2724
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0021396314259369e-05,
+      "loss": 0.7585,
+      "step": 2725
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0015283087325377e-05,
+      "loss": 0.762,
+      "step": 2726
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0009169854679848e-05,
+      "loss": 0.7744,
+      "step": 2727
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0003056618607388e-05,
+      "loss": 0.7954,
+      "step": 2728
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.996943381392616e-06,
+      "loss": 0.7754,
+      "step": 2729
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.990830145320154e-06,
+      "loss": 0.7734,
+      "step": 2730
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.984716912674624e-06,
+      "loss": 0.7627,
+      "step": 2731
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.978603685740635e-06,
+      "loss": 0.7993,
+      "step": 2732
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.972490466802806e-06,
+      "loss": 0.7891,
+      "step": 2733
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96637725814575e-06,
+      "loss": 0.7832,
+      "step": 2734
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960264062054072e-06,
+      "loss": 0.7927,
+      "step": 2735
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.954150880812377e-06,
+      "loss": 0.7632,
+      "step": 2736
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.94803771670527e-06,
+      "loss": 0.7668,
+      "step": 2737
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.941924572017329e-06,
+      "loss": 0.7756,
+      "step": 2738
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.935811449033145e-06,
+      "loss": 0.7871,
+      "step": 2739
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.929698350037296e-06,
+      "loss": 0.7795,
+      "step": 2740
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.923585277314345e-06,
+      "loss": 0.7881,
+      "step": 2741
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.917472233148852e-06,
+      "loss": 0.7649,
+      "step": 2742
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.911359219825366e-06,
+      "loss": 0.752,
+      "step": 2743
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.905246239628416e-06,
+      "loss": 0.8037,
+      "step": 2744
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.899133294842528e-06,
+      "loss": 0.7812,
+      "step": 2745
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.893020387752211e-06,
+      "loss": 0.7773,
+      "step": 2746
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.886907520641963e-06,
+      "loss": 0.8005,
+      "step": 2747
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88079469579626e-06,
+      "loss": 0.8152,
+      "step": 2748
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.87468191549957e-06,
+      "loss": 0.7852,
+      "step": 2749
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.868569182036345e-06,
+      "loss": 0.8074,
+      "step": 2750
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.862456497691004e-06,
+      "loss": 0.7649,
+      "step": 2751
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.856343864747967e-06,
+      "loss": 0.8145,
+      "step": 2752
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.85023128549163e-06,
+      "loss": 0.7729,
+      "step": 2753
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.844118762206357e-06,
+      "loss": 0.804,
+      "step": 2754
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.838006297176505e-06,
+      "loss": 0.7844,
+      "step": 2755
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.831893892686406e-06,
+      "loss": 0.8,
+      "step": 2756
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.825781551020363e-06,
+      "loss": 0.803,
+      "step": 2757
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.819669274462662e-06,
+      "loss": 0.7832,
+      "step": 2758
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.813557065297568e-06,
+      "loss": 0.7678,
+      "step": 2759
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.807444925809306e-06,
+      "loss": 0.761,
+      "step": 2760
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.801332858282092e-06,
+      "loss": 0.8213,
+      "step": 2761
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.795220865000102e-06,
+      "loss": 0.7415,
+      "step": 2762
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.789108948247497e-06,
+      "loss": 0.7632,
+      "step": 2763
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.782997110308394e-06,
+      "loss": 0.7922,
+      "step": 2764
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.776885353466895e-06,
+      "loss": 0.8291,
+      "step": 2765
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.770773680007068e-06,
+      "loss": 0.8057,
+      "step": 2766
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.764662092212939e-06,
+      "loss": 0.7399,
+      "step": 2767
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.758550592368514e-06,
+      "loss": 0.7681,
+      "step": 2768
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.752439182757765e-06,
+      "loss": 0.8071,
+      "step": 2769
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.746327865664623e-06,
+      "loss": 0.7849,
+      "step": 2770
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.740216643372993e-06,
+      "loss": 0.7966,
+      "step": 2771
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.734105518166745e-06,
+      "loss": 0.7861,
+      "step": 2772
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.727994492329697e-06,
+      "loss": 0.7927,
+      "step": 2773
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72188356814565e-06,
+      "loss": 0.7915,
+      "step": 2774
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.715772747898359e-06,
+      "loss": 0.7712,
+      "step": 2775
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.709662033871532e-06,
+      "loss": 0.7781,
+      "step": 2776
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.70355142834885e-06,
+      "loss": 0.7925,
+      "step": 2777
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.697440933613947e-06,
+      "loss": 0.7556,
+      "step": 2778
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69133055195042e-06,
+      "loss": 0.7241,
+      "step": 2779
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685220285641815e-06,
+      "loss": 0.8157,
+      "step": 2780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.679110136971646e-06,
+      "loss": 0.7812,
+      "step": 2781
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.673000108223377e-06,
+      "loss": 0.7825,
+      "step": 2782
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.666890201680423e-06,
+      "loss": 0.7979,
+      "step": 2783
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.66078041962616e-06,
+      "loss": 0.7742,
+      "step": 2784
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.654670764343921e-06,
+      "loss": 0.7727,
+      "step": 2785
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.648561238116978e-06,
+      "loss": 0.8083,
+      "step": 2786
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.642451843228567e-06,
+      "loss": 0.7983,
+      "step": 2787
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.636342581961875e-06,
+      "loss": 0.7444,
+      "step": 2788
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.630233456600027e-06,
+      "loss": 0.8062,
+      "step": 2789
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.624124469426107e-06,
+      "loss": 0.7568,
+      "step": 2790
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.61801562272315e-06,
+      "loss": 0.8125,
+      "step": 2791
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.611906918774129e-06,
+      "loss": 0.7881,
+      "step": 2792
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.60579835986197e-06,
+      "loss": 0.8101,
+      "step": 2793
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.599689948269545e-06,
+      "loss": 0.772,
+      "step": 2794
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.593581686279672e-06,
+      "loss": 0.7544,
+      "step": 2795
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.587473576175103e-06,
+      "loss": 0.7878,
+      "step": 2796
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.581365620238544e-06,
+      "loss": 0.7908,
+      "step": 2797
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.575257820752646e-06,
+      "loss": 0.7966,
+      "step": 2798
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.569150179999985e-06,
+      "loss": 0.771,
+      "step": 2799
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.563042700263094e-06,
+      "loss": 0.7551,
+      "step": 2800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.556935383824443e-06,
+      "loss": 0.7405,
+      "step": 2801
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.55082823296643e-06,
+      "loss": 0.7688,
+      "step": 2802
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.544721249971408e-06,
+      "loss": 0.7822,
+      "step": 2803
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.538614437121658e-06,
+      "loss": 0.7544,
+      "step": 2804
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.532507796699388e-06,
+      "loss": 0.7634,
+      "step": 2805
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.526401330986759e-06,
+      "loss": 0.8027,
+      "step": 2806
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.52029504226586e-06,
+      "loss": 0.8032,
+      "step": 2807
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.514188932818711e-06,
+      "loss": 0.754,
+      "step": 2808
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.508083004927265e-06,
+      "loss": 0.8071,
+      "step": 2809
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.501977260873412e-06,
+      "loss": 0.7603,
+      "step": 2810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.495871702938974e-06,
+      "loss": 0.7974,
+      "step": 2811
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.489766333405692e-06,
+      "loss": 0.7847,
+      "step": 2812
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.483661154555248e-06,
+      "loss": 0.8118,
+      "step": 2813
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.47755616866925e-06,
+      "loss": 0.8091,
+      "step": 2814
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.471451378029232e-06,
+      "loss": 0.7568,
+      "step": 2815
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.465346784916658e-06,
+      "loss": 0.7771,
+      "step": 2816
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.459242391612919e-06,
+      "loss": 0.7507,
+      "step": 2817
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.453138200399318e-06,
+      "loss": 0.748,
+      "step": 2818
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.447034213557102e-06,
+      "loss": 0.8123,
+      "step": 2819
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.440930433367436e-06,
+      "loss": 0.7676,
+      "step": 2820
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.434826862111392e-06,
+      "loss": 0.813,
+      "step": 2821
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.428723502069987e-06,
+      "loss": 0.7788,
+      "step": 2822
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.422620355524144e-06,
+      "loss": 0.7776,
+      "step": 2823
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.416517424754712e-06,
+      "loss": 0.759,
+      "step": 2824
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.410414712042459e-06,
+      "loss": 0.7944,
+      "step": 2825
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.404312219668068e-06,
+      "loss": 0.7627,
+      "step": 2826
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.39820994991215e-06,
+      "loss": 0.7454,
+      "step": 2827
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.392107905055213e-06,
+      "loss": 0.8091,
+      "step": 2828
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.386006087377699e-06,
+      "loss": 0.7739,
+      "step": 2829
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.379904499159962e-06,
+      "loss": 0.7805,
+      "step": 2830
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.373803142682263e-06,
+      "loss": 0.7634,
+      "step": 2831
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.367702020224782e-06,
+      "loss": 0.7634,
+      "step": 2832
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.361601134067616e-06,
+      "loss": 0.8088,
+      "step": 2833
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.355500486490756e-06,
+      "loss": 0.7417,
+      "step": 2834
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.349400079774124e-06,
+      "loss": 0.7561,
+      "step": 2835
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.343299916197544e-06,
+      "loss": 0.7512,
+      "step": 2836
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.337199998040744e-06,
+      "loss": 0.7708,
+      "step": 2837
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.33110032758337e-06,
+      "loss": 0.7922,
+      "step": 2838
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.325000907104972e-06,
+      "loss": 0.7842,
+      "step": 2839
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.318901738885e-06,
+      "loss": 0.7505,
+      "step": 2840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.312802825202817e-06,
+      "loss": 0.8108,
+      "step": 2841
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30670416833769e-06,
+      "loss": 0.7827,
+      "step": 2842
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30060577056879e-06,
+      "loss": 0.771,
+      "step": 2843
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.294507634175185e-06,
+      "loss": 0.792,
+      "step": 2844
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.288409761435854e-06,
+      "loss": 0.7722,
+      "step": 2845
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.282312154629675e-06,
+      "loss": 0.7607,
+      "step": 2846
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.27621481603542e-06,
+      "loss": 0.8362,
+      "step": 2847
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.270117747931772e-06,
+      "loss": 0.7817,
+      "step": 2848
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.26402095259731e-06,
+      "loss": 0.78,
+      "step": 2849
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.257924432310497e-06,
+      "loss": 0.8005,
+      "step": 2850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.251828189349713e-06,
+      "loss": 0.7825,
+      "step": 2851
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.245732225993223e-06,
+      "loss": 0.7815,
+      "step": 2852
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.23963654451919e-06,
+      "loss": 0.7905,
+      "step": 2853
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.233541147205672e-06,
+      "loss": 0.8103,
+      "step": 2854
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.227446036330627e-06,
+      "loss": 0.7976,
+      "step": 2855
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.221351214171888e-06,
+      "loss": 0.782,
+      "step": 2856
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.215256683007198e-06,
+      "loss": 0.8096,
+      "step": 2857
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.209162445114187e-06,
+      "loss": 0.7874,
+      "step": 2858
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.203068502770371e-06,
+      "loss": 0.7695,
+      "step": 2859
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.196974858253158e-06,
+      "loss": 0.7576,
+      "step": 2860
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.190881513839844e-06,
+      "loss": 0.7974,
+      "step": 2861
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.184788471807619e-06,
+      "loss": 0.7893,
+      "step": 2862
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.178695734433544e-06,
+      "loss": 0.7837,
+      "step": 2863
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.172603303994584e-06,
+      "loss": 0.7766,
+      "step": 2864
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.166511182767585e-06,
+      "loss": 0.7522,
+      "step": 2865
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.160419373029266e-06,
+      "loss": 0.7747,
+      "step": 2866
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.154327877056242e-06,
+      "loss": 0.7482,
+      "step": 2867
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.148236697125009e-06,
+      "loss": 0.7961,
+      "step": 2868
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.142145835511937e-06,
+      "loss": 0.7937,
+      "step": 2869
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.136055294493285e-06,
+      "loss": 0.7615,
+      "step": 2870
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.129965076345197e-06,
+      "loss": 0.7151,
+      "step": 2871
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.123875183343677e-06,
+      "loss": 0.7778,
+      "step": 2872
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.117785617764627e-06,
+      "loss": 0.7549,
+      "step": 2873
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.111696381883819e-06,
+      "loss": 0.7661,
+      "step": 2874
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.105607477976901e-06,
+      "loss": 0.7717,
+      "step": 2875
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.099518908319397e-06,
+      "loss": 0.8083,
+      "step": 2876
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.093430675186709e-06,
+      "loss": 0.793,
+      "step": 2877
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.087342780854114e-06,
+      "loss": 0.7683,
+      "step": 2878
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.081255227596751e-06,
+      "loss": 0.7708,
+      "step": 2879
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.075168017689646e-06,
+      "loss": 0.7913,
+      "step": 2880
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.069081153407695e-06,
+      "loss": 0.7318,
+      "step": 2881
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.062994637025651e-06,
+      "loss": 0.7546,
+      "step": 2882
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.056908470818153e-06,
+      "loss": 0.8035,
+      "step": 2883
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.050822657059705e-06,
+      "loss": 0.7522,
+      "step": 2884
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.044737198024673e-06,
+      "loss": 0.7275,
+      "step": 2885
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.038652095987294e-06,
+      "loss": 0.7437,
+      "step": 2886
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.032567353221679e-06,
+      "loss": 0.7793,
+      "step": 2887
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.026482972001786e-06,
+      "loss": 0.8088,
+      "step": 2888
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.02039895460146e-06,
+      "loss": 0.7834,
+      "step": 2889
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.014315303294396e-06,
+      "loss": 0.7954,
+      "step": 2890
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.008232020354158e-06,
+      "loss": 0.7729,
+      "step": 2891
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.002149108054166e-06,
+      "loss": 0.7695,
+      "step": 2892
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.99606656866771e-06,
+      "loss": 0.7871,
+      "step": 2893
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.98998440446794e-06,
+      "loss": 0.8037,
+      "step": 2894
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.983902617727851e-06,
+      "loss": 0.7766,
+      "step": 2895
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.977821210720318e-06,
+      "loss": 0.7698,
+      "step": 2896
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.971740185718062e-06,
+      "loss": 0.7734,
+      "step": 2897
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96565954499366e-06,
+      "loss": 0.7678,
+      "step": 2898
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.959579290819554e-06,
+      "loss": 0.7996,
+      "step": 2899
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.953499425468036e-06,
+      "loss": 0.7783,
+      "step": 2900
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.947419951211248e-06,
+      "loss": 0.8337,
+      "step": 2901
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.941340870321195e-06,
+      "loss": 0.7773,
+      "step": 2902
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.935262185069731e-06,
+      "loss": 0.7803,
+      "step": 2903
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.929183897728561e-06,
+      "loss": 0.7876,
+      "step": 2904
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.923106010569242e-06,
+      "loss": 0.7964,
+      "step": 2905
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.917028525863183e-06,
+      "loss": 0.7739,
+      "step": 2906
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.910951445881646e-06,
+      "loss": 0.7676,
+      "step": 2907
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.904874772895727e-06,
+      "loss": 0.7661,
+      "step": 2908
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.898798509176387e-06,
+      "loss": 0.8003,
+      "step": 2909
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.89272265699443e-06,
+      "loss": 0.7627,
+      "step": 2910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.886647218620494e-06,
+      "loss": 0.7678,
+      "step": 2911
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.880572196325079e-06,
+      "loss": 0.7603,
+      "step": 2912
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.87449759237852e-06,
+      "loss": 0.7957,
+      "step": 2913
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.868423409050998e-06,
+      "loss": 0.7712,
+      "step": 2914
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.862349648612536e-06,
+      "loss": 0.7578,
+      "step": 2915
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.856276313333004e-06,
+      "loss": 0.7649,
+      "step": 2916
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.850203405482102e-06,
+      "loss": 0.7725,
+      "step": 2917
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.84413092732938e-06,
+      "loss": 0.739,
+      "step": 2918
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.838058881144226e-06,
+      "loss": 0.7793,
+      "step": 2919
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.831987269195863e-06,
+      "loss": 0.7446,
+      "step": 2920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.825916093753352e-06,
+      "loss": 0.7903,
+      "step": 2921
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.819845357085599e-06,
+      "loss": 0.7864,
+      "step": 2922
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.813775061461338e-06,
+      "loss": 0.8008,
+      "step": 2923
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.807705209149132e-06,
+      "loss": 0.7949,
+      "step": 2924
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.801635802417394e-06,
+      "loss": 0.8132,
+      "step": 2925
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79556684353436e-06,
+      "loss": 0.7751,
+      "step": 2926
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.789498334768101e-06,
+      "loss": 0.7651,
+      "step": 2927
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.78343027838652e-06,
+      "loss": 0.8052,
+      "step": 2928
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.777362676657355e-06,
+      "loss": 0.8284,
+      "step": 2929
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.771295531848162e-06,
+      "loss": 0.7786,
+      "step": 2930
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.765228846226342e-06,
+      "loss": 0.7539,
+      "step": 2931
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.759162622059115e-06,
+      "loss": 0.7888,
+      "step": 2932
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.753096861613525e-06,
+      "loss": 0.801,
+      "step": 2933
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.747031567156452e-06,
+      "loss": 0.7493,
+      "step": 2934
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.7409667409546e-06,
+      "loss": 0.7634,
+      "step": 2935
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.73490238527449e-06,
+      "loss": 0.7937,
+      "step": 2936
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.728838502382477e-06,
+      "loss": 0.7979,
+      "step": 2937
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.722775094544734e-06,
+      "loss": 0.7637,
+      "step": 2938
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.716712164027263e-06,
+      "loss": 0.8184,
+      "step": 2939
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.710649713095873e-06,
+      "loss": 0.7976,
+      "step": 2940
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.704587744016206e-06,
+      "loss": 0.7607,
+      "step": 2941
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.698526259053727e-06,
+      "loss": 0.8142,
+      "step": 2942
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.692465260473707e-06,
+      "loss": 0.8079,
+      "step": 2943
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.686404750541246e-06,
+      "loss": 0.7903,
+      "step": 2944
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68034473152126e-06,
+      "loss": 0.7908,
+      "step": 2945
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.674285205678471e-06,
+      "loss": 0.7864,
+      "step": 2946
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.668226175277431e-06,
+      "loss": 0.7278,
+      "step": 2947
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.662167642582502e-06,
+      "loss": 0.7888,
+      "step": 2948
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.656109609857855e-06,
+      "loss": 0.7739,
+      "step": 2949
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.650052079367478e-06,
+      "loss": 0.7656,
+      "step": 2950
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.643995053375178e-06,
+      "loss": 0.7649,
+      "step": 2951
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.637938534144558e-06,
+      "loss": 0.7981,
+      "step": 2952
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.631882523939043e-06,
+      "loss": 0.7844,
+      "step": 2953
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.625827025021867e-06,
+      "loss": 0.7764,
+      "step": 2954
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.619772039656072e-06,
+      "loss": 0.7239,
+      "step": 2955
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.613717570104502e-06,
+      "loss": 0.8096,
+      "step": 2956
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.607663618629818e-06,
+      "loss": 0.8037,
+      "step": 2957
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.601610187494485e-06,
+      "loss": 0.7664,
+      "step": 2958
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.595557278960762e-06,
+      "loss": 0.7744,
+      "step": 2959
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.589504895290727e-06,
+      "loss": 0.7737,
+      "step": 2960
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.583453038746265e-06,
+      "loss": 0.8196,
+      "step": 2961
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.577401711589041e-06,
+      "loss": 0.7844,
+      "step": 2962
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.571350916080544e-06,
+      "loss": 0.7849,
+      "step": 2963
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.565300654482061e-06,
+      "loss": 0.7856,
+      "step": 2964
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.559250929054667e-06,
+      "loss": 0.8174,
+      "step": 2965
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.553201742059254e-06,
+      "loss": 0.8071,
+      "step": 2966
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.547153095756504e-06,
+      "loss": 0.761,
+      "step": 2967
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.541104992406887e-06,
+      "loss": 0.7681,
+      "step": 2968
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53505743427069e-06,
+      "loss": 0.7285,
+      "step": 2969
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52901042360798e-06,
+      "loss": 0.7817,
+      "step": 2970
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.522963962678635e-06,
+      "loss": 0.7634,
+      "step": 2971
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.516918053742307e-06,
+      "loss": 0.7837,
+      "step": 2972
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.51087269905846e-06,
+      "loss": 0.752,
+      "step": 2973
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.504827900886347e-06,
+      "loss": 0.7329,
+      "step": 2974
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.498783661485e-06,
+      "loss": 0.7769,
+      "step": 2975
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.49273998311326e-06,
+      "loss": 0.7864,
+      "step": 2976
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.486696868029747e-06,
+      "loss": 0.7749,
+      "step": 2977
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.480654318492876e-06,
+      "loss": 0.8357,
+      "step": 2978
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.474612336760846e-06,
+      "loss": 0.833,
+      "step": 2979
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.468570925091653e-06,
+      "loss": 0.77,
+      "step": 2980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.462530085743066e-06,
+      "loss": 0.7886,
+      "step": 2981
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.456489820972648e-06,
+      "loss": 0.8047,
+      "step": 2982
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.450450133037754e-06,
+      "loss": 0.7925,
+      "step": 2983
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.444411024195506e-06,
+      "loss": 0.7551,
+      "step": 2984
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.438372496702826e-06,
+      "loss": 0.7888,
+      "step": 2985
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43233455281641e-06,
+      "loss": 0.8032,
+      "step": 2986
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.426297194792742e-06,
+      "loss": 0.7656,
+      "step": 2987
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.420260424888081e-06,
+      "loss": 0.7544,
+      "step": 2988
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.414224245358467e-06,
+      "loss": 0.7917,
+      "step": 2989
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.408188658459728e-06,
+      "loss": 0.8071,
+      "step": 2990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.402153666447455e-06,
+      "loss": 0.7825,
+      "step": 2991
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.396119271577026e-06,
+      "loss": 0.7722,
+      "step": 2992
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.390085476103603e-06,
+      "loss": 0.749,
+      "step": 2993
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.384052282282109e-06,
+      "loss": 0.7908,
+      "step": 2994
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.378019692367249e-06,
+      "loss": 0.7788,
+      "step": 2995
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.37198770861351e-06,
+      "loss": 0.7588,
+      "step": 2996
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.365956333275136e-06,
+      "loss": 0.7827,
+      "step": 2997
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.359925568606156e-06,
+      "loss": 0.8008,
+      "step": 2998
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35389541686037e-06,
+      "loss": 0.8384,
+      "step": 2999
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.347865880291341e-06,
+      "loss": 0.7744,
+      "step": 3000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.341836961152413e-06,
+      "loss": 0.8076,
+      "step": 3001
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33580866169669e-06,
+      "loss": 0.7957,
+      "step": 3002
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.329780984177054e-06,
+      "loss": 0.7274,
+      "step": 3003
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.323753930846139e-06,
+      "loss": 0.7725,
+      "step": 3004
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.317727503956362e-06,
+      "loss": 0.7573,
+      "step": 3005
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.311701705759903e-06,
+      "loss": 0.7798,
+      "step": 3006
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.305676538508693e-06,
+      "loss": 0.8032,
+      "step": 3007
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.299652004454441e-06,
+      "loss": 0.7458,
+      "step": 3008
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.293628105848623e-06,
+      "loss": 0.6975,
+      "step": 3009
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.287604844942462e-06,
+      "loss": 0.7629,
+      "step": 3010
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.281582223986954e-06,
+      "loss": 0.7676,
+      "step": 3011
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.27556024523286e-06,
+      "loss": 0.752,
+      "step": 3012
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.269538910930682e-06,
+      "loss": 0.7866,
+      "step": 3013
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.7703,
+      "step": 3014
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.257498184682942e-06,
+      "loss": 0.7617,
+      "step": 3015
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.2514787972372e-06,
+      "loss": 0.7761,
+      "step": 3016
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.245460063243017e-06,
+      "loss": 0.8098,
+      "step": 3017
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.239441984949696e-06,
+      "loss": 0.7737,
+      "step": 3018
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.2334245646063e-06,
+      "loss": 0.769,
+      "step": 3019
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.227407804461626e-06,
+      "loss": 0.8008,
+      "step": 3020
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.221391706764246e-06,
+      "loss": 0.782,
+      "step": 3021
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.215376273762475e-06,
+      "loss": 0.7815,
+      "step": 3022
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.209361507704379e-06,
+      "loss": 0.8247,
+      "step": 3023
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.203347410837778e-06,
+      "loss": 0.7646,
+      "step": 3024
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.197333985410244e-06,
+      "loss": 0.7891,
+      "step": 3025
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.191321233669085e-06,
+      "loss": 0.7795,
+      "step": 3026
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.185309157861374e-06,
+      "loss": 0.7175,
+      "step": 3027
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.179297760233927e-06,
+      "loss": 0.7888,
+      "step": 3028
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.173287043033292e-06,
+      "loss": 0.7808,
+      "step": 3029
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.167277008505782e-06,
+      "loss": 0.7854,
+      "step": 3030
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.161267658897452e-06,
+      "loss": 0.7625,
+      "step": 3031
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.155258996454085e-06,
+      "loss": 0.7629,
+      "step": 3032
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.149251023421226e-06,
+      "loss": 0.7522,
+      "step": 3033
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.143243742044153e-06,
+      "loss": 0.7209,
+      "step": 3034
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13723715456789e-06,
+      "loss": 0.7378,
+      "step": 3035
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.131231263237194e-06,
+      "loss": 0.8052,
+      "step": 3036
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12522607029657e-06,
+      "loss": 0.7859,
+      "step": 3037
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.119221577990258e-06,
+      "loss": 0.7566,
+      "step": 3038
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.113217788562236e-06,
+      "loss": 0.8052,
+      "step": 3039
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10721470425622e-06,
+      "loss": 0.7656,
+      "step": 3040
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.101212327315671e-06,
+      "loss": 0.7595,
+      "step": 3041
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.095210659983765e-06,
+      "loss": 0.7874,
+      "step": 3042
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.089209704503432e-06,
+      "loss": 0.7737,
+      "step": 3043
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.08320946311733e-06,
+      "loss": 0.7612,
+      "step": 3044
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.077209938067843e-06,
+      "loss": 0.7961,
+      "step": 3045
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.0712111315971e-06,
+      "loss": 0.7998,
+      "step": 3046
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.065213045946958e-06,
+      "loss": 0.7715,
+      "step": 3047
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.059215683358992e-06,
+      "loss": 0.7993,
+      "step": 3048
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.053219046074521e-06,
+      "loss": 0.7769,
+      "step": 3049
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04722313633459e-06,
+      "loss": 0.7974,
+      "step": 3050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.041227956379975e-06,
+      "loss": 0.7498,
+      "step": 3051
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.035233508451163e-06,
+      "loss": 0.7927,
+      "step": 3052
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.029239794788386e-06,
+      "loss": 0.8271,
+      "step": 3053
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.023246817631597e-06,
+      "loss": 0.7859,
+      "step": 3054
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.017254579220468e-06,
+      "loss": 0.7793,
+      "step": 3055
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0112630817944e-06,
+      "loss": 0.7854,
+      "step": 3056
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.005272327592518e-06,
+      "loss": 0.7625,
+      "step": 3057
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99928231885366e-06,
+      "loss": 0.7551,
+      "step": 3058
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.993293057816397e-06,
+      "loss": 0.7969,
+      "step": 3059
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.987304546719015e-06,
+      "loss": 0.783,
+      "step": 3060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98131678779952e-06,
+      "loss": 0.7654,
+      "step": 3061
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.975329783295638e-06,
+      "loss": 0.8022,
+      "step": 3062
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.969343535444818e-06,
+      "loss": 0.7925,
+      "step": 3063
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.96335804648421e-06,
+      "loss": 0.8083,
+      "step": 3064
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.957373318650697e-06,
+      "loss": 0.8328,
+      "step": 3065
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.951389354180872e-06,
+      "loss": 0.7922,
+      "step": 3066
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.945406155311041e-06,
+      "loss": 0.76,
+      "step": 3067
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.939423724277226e-06,
+      "loss": 0.7764,
+      "step": 3068
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.933442063315161e-06,
+      "loss": 0.7725,
+      "step": 3069
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.927461174660298e-06,
+      "loss": 0.77,
+      "step": 3070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.921481060547784e-06,
+      "loss": 0.7312,
+      "step": 3071
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.915501723212494e-06,
+      "loss": 0.7549,
+      "step": 3072
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.909523164889011e-06,
+      "loss": 0.7979,
+      "step": 3073
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.90354538781161e-06,
+      "loss": 0.7527,
+      "step": 3074
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.897568394214292e-06,
+      "loss": 0.7776,
+      "step": 3075
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.891592186330764e-06,
+      "loss": 0.7834,
+      "step": 3076
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.885616766394426e-06,
+      "loss": 0.7749,
+      "step": 3077
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.8796421366384e-06,
+      "loss": 0.7964,
+      "step": 3078
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.873668299295504e-06,
+      "loss": 0.8076,
+      "step": 3079
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.867695256598254e-06,
+      "loss": 0.7656,
+      "step": 3080
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86172301077888e-06,
+      "loss": 0.7976,
+      "step": 3081
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.855751564069309e-06,
+      "loss": 0.7708,
+      "step": 3082
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.849780918701173e-06,
+      "loss": 0.7876,
+      "step": 3083
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.8438110769058e-06,
+      "loss": 0.7444,
+      "step": 3084
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.837842040914219e-06,
+      "loss": 0.8091,
+      "step": 3085
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.831873812957162e-06,
+      "loss": 0.7532,
+      "step": 3086
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.825906395265049e-06,
+      "loss": 0.7402,
+      "step": 3087
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.819939790068007e-06,
+      "loss": 0.7944,
+      "step": 3088
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81397399959586e-06,
+      "loss": 0.7898,
+      "step": 3089
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.808009026078113e-06,
+      "loss": 0.7778,
+      "step": 3090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.802044871743986e-06,
+      "loss": 0.7417,
+      "step": 3091
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.796081538822384e-06,
+      "loss": 0.8101,
+      "step": 3092
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.790119029541894e-06,
+      "loss": 0.7573,
+      "step": 3093
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.78415734613081e-06,
+      "loss": 0.7644,
+      "step": 3094
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.778196490817121e-06,
+      "loss": 0.7954,
+      "step": 3095
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.772236465828485e-06,
+      "loss": 0.7649,
+      "step": 3096
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.766277273392268e-06,
+      "loss": 0.7856,
+      "step": 3097
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76031891573552e-06,
+      "loss": 0.7634,
+      "step": 3098
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.754361395084982e-06,
+      "loss": 0.7712,
+      "step": 3099
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.748404713667072e-06,
+      "loss": 0.7766,
+      "step": 3100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.742448873707907e-06,
+      "loss": 0.7281,
+      "step": 3101
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.736493877433285e-06,
+      "loss": 0.7942,
+      "step": 3102
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73053972706868e-06,
+      "loss": 0.7896,
+      "step": 3103
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.724586424839264e-06,
+      "loss": 0.814,
+      "step": 3104
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.718633972969885e-06,
+      "loss": 0.7512,
+      "step": 3105
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.712682373685067e-06,
+      "loss": 0.752,
+      "step": 3106
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70673162920903e-06,
+      "loss": 0.7576,
+      "step": 3107
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700781741765667e-06,
+      "loss": 0.7712,
+      "step": 3108
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.694832713578541e-06,
+      "loss": 0.7563,
+      "step": 3109
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.688884546870908e-06,
+      "loss": 0.8125,
+      "step": 3110
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.682937243865702e-06,
+      "loss": 0.7214,
+      "step": 3111
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.676990806785523e-06,
+      "loss": 0.8047,
+      "step": 3112
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.671045237852657e-06,
+      "loss": 0.8027,
+      "step": 3113
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.66510053928906e-06,
+      "loss": 0.7815,
+      "step": 3114
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.659156713316371e-06,
+      "loss": 0.803,
+      "step": 3115
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65321376215589e-06,
+      "loss": 0.7759,
+      "step": 3116
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.647271688028598e-06,
+      "loss": 0.7429,
+      "step": 3117
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.641330493155153e-06,
+      "loss": 0.7681,
+      "step": 3118
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.635390179755871e-06,
+      "loss": 0.7883,
+      "step": 3119
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.629450750050751e-06,
+      "loss": 0.7629,
+      "step": 3120
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.623512206259457e-06,
+      "loss": 0.8052,
+      "step": 3121
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.617574550601319e-06,
+      "loss": 0.7852,
+      "step": 3122
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6116377852953395e-06,
+      "loss": 0.7723,
+      "step": 3123
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.605701912560193e-06,
+      "loss": 0.7483,
+      "step": 3124
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.599766934614202e-06,
+      "loss": 0.7593,
+      "step": 3125
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.593832853675375e-06,
+      "loss": 0.7439,
+      "step": 3126
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5878996719613765e-06,
+      "loss": 0.751,
+      "step": 3127
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5819673916895335e-06,
+      "loss": 0.8032,
+      "step": 3128
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.576036015076838e-06,
+      "loss": 0.7783,
+      "step": 3129
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.570105544339945e-06,
+      "loss": 0.7529,
+      "step": 3130
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.564175981695176e-06,
+      "loss": 0.7607,
+      "step": 3131
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.558247329358498e-06,
+      "loss": 0.7456,
+      "step": 3132
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.552319589545552e-06,
+      "loss": 0.7495,
+      "step": 3133
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5463927644716354e-06,
+      "loss": 0.7849,
+      "step": 3134
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.540466856351698e-06,
+      "loss": 0.7882,
+      "step": 3135
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.534541867400351e-06,
+      "loss": 0.7795,
+      "step": 3136
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.528617799831867e-06,
+      "loss": 0.7637,
+      "step": 3137
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.522694655860161e-06,
+      "loss": 0.8235,
+      "step": 3138
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.516772437698814e-06,
+      "loss": 0.7693,
+      "step": 3139
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.510851147561061e-06,
+      "loss": 0.7922,
+      "step": 3140
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.504930787659781e-06,
+      "loss": 0.7517,
+      "step": 3141
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.499011360207516e-06,
+      "loss": 0.7842,
+      "step": 3142
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.493092867416454e-06,
+      "loss": 0.7749,
+      "step": 3143
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4871753114984314e-06,
+      "loss": 0.8013,
+      "step": 3144
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4812586946649405e-06,
+      "loss": 0.7759,
+      "step": 3145
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.475343019127119e-06,
+      "loss": 0.7788,
+      "step": 3146
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.469428287095758e-06,
+      "loss": 0.7646,
+      "step": 3147
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.463514500781284e-06,
+      "loss": 0.8184,
+      "step": 3148
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.45760166239378e-06,
+      "loss": 0.7537,
+      "step": 3149
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.451689774142977e-06,
+      "loss": 0.7632,
+      "step": 3150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.44577883823824e-06,
+      "loss": 0.7493,
+      "step": 3151
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.439868856888588e-06,
+      "loss": 0.7778,
+      "step": 3152
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.433959832302684e-06,
+      "loss": 0.7844,
+      "step": 3153
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.428051766688819e-06,
+      "loss": 0.7744,
+      "step": 3154
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.42214466225494e-06,
+      "loss": 0.7805,
+      "step": 3155
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.416238521208634e-06,
+      "loss": 0.745,
+      "step": 3156
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.410333345757121e-06,
+      "loss": 0.8069,
+      "step": 3157
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.404429138107263e-06,
+      "loss": 0.7688,
+      "step": 3158
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.398525900465567e-06,
+      "loss": 0.7571,
+      "step": 3159
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.392623635038164e-06,
+      "loss": 0.7803,
+      "step": 3160
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.386722344030831e-06,
+      "loss": 0.7832,
+      "step": 3161
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.38082202964898e-06,
+      "loss": 0.7554,
+      "step": 3162
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.374922694097662e-06,
+      "loss": 0.7776,
+      "step": 3163
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.369024339581548e-06,
+      "loss": 0.8032,
+      "step": 3164
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.363126968304954e-06,
+      "loss": 0.8052,
+      "step": 3165
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.35723058247183e-06,
+      "loss": 0.7561,
+      "step": 3166
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.351335184285752e-06,
+      "loss": 0.7705,
+      "step": 3167
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.345440775949926e-06,
+      "loss": 0.804,
+      "step": 3168
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.339547359667199e-06,
+      "loss": 0.7722,
+      "step": 3169
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.333654937640027e-06,
+      "loss": 0.7654,
+      "step": 3170
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.327763512070514e-06,
+      "loss": 0.7722,
+      "step": 3171
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.321873085160385e-06,
+      "loss": 0.739,
+      "step": 3172
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.315983659110986e-06,
+      "loss": 0.7886,
+      "step": 3173
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.310095236123296e-06,
+      "loss": 0.7612,
+      "step": 3174
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.304207818397921e-06,
+      "loss": 0.7944,
+      "step": 3175
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.29832140813508e-06,
+      "loss": 0.7549,
+      "step": 3176
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2924360075346266e-06,
+      "loss": 0.7427,
+      "step": 3177
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.286551618796034e-06,
+      "loss": 0.7539,
+      "step": 3178
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.280668244118397e-06,
+      "loss": 0.7896,
+      "step": 3179
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2747858857004265e-06,
+      "loss": 0.7776,
+      "step": 3180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.268904545740461e-06,
+      "loss": 0.7739,
+      "step": 3181
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.26302422643646e-06,
+      "loss": 0.7461,
+      "step": 3182
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.257144929985989e-06,
+      "loss": 0.7249,
+      "step": 3183
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.251266658586241e-06,
+      "loss": 0.7769,
+      "step": 3184
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.245389414434031e-06,
+      "loss": 0.7441,
+      "step": 3185
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.239513199725774e-06,
+      "loss": 0.7251,
+      "step": 3186
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.233638016657513e-06,
+      "loss": 0.783,
+      "step": 3187
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.227763867424904e-06,
+      "loss": 0.7502,
+      "step": 3188
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.221890754223213e-06,
+      "loss": 0.7104,
+      "step": 3189
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.216018679247319e-06,
+      "loss": 0.7842,
+      "step": 3190
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.21014764469172e-06,
+      "loss": 0.7852,
+      "step": 3191
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.204277652750513e-06,
+      "loss": 0.7671,
+      "step": 3192
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.198408705617413e-06,
+      "loss": 0.7427,
+      "step": 3193
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.192540805485746e-06,
+      "loss": 0.7429,
+      "step": 3194
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.186673954548444e-06,
+      "loss": 0.7434,
+      "step": 3195
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.180808154998047e-06,
+      "loss": 0.803,
+      "step": 3196
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1749434090267e-06,
+      "loss": 0.749,
+      "step": 3197
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.169079718826162e-06,
+      "loss": 0.8005,
+      "step": 3198
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.163217086587784e-06,
+      "loss": 0.801,
+      "step": 3199
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.157355514502534e-06,
+      "loss": 0.748,
+      "step": 3200
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.151495004760981e-06,
+      "loss": 0.7603,
+      "step": 3201
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.145635559553291e-06,
+      "loss": 0.7791,
+      "step": 3202
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.139777181069238e-06,
+      "loss": 0.7693,
+      "step": 3203
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.133919871498203e-06,
+      "loss": 0.7954,
+      "step": 3204
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.128063633029147e-06,
+      "loss": 0.7761,
+      "step": 3205
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.122208467850653e-06,
+      "loss": 0.7861,
+      "step": 3206
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.116354378150895e-06,
+      "loss": 0.7434,
+      "step": 3207
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.110501366117637e-06,
+      "loss": 0.8018,
+      "step": 3208
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.104649433938251e-06,
+      "loss": 0.7412,
+      "step": 3209
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.098798583799702e-06,
+      "loss": 0.7783,
+      "step": 3210
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.092948817888554e-06,
+      "loss": 0.7988,
+      "step": 3211
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0871001383909546e-06,
+      "loss": 0.8088,
+      "step": 3212
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0812525474926574e-06,
+      "loss": 0.7803,
+      "step": 3213
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.075406047379009e-06,
+      "loss": 0.7124,
+      "step": 3214
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0695606402349335e-06,
+      "loss": 0.7356,
+      "step": 3215
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.063716328244965e-06,
+      "loss": 0.783,
+      "step": 3216
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.057873113593219e-06,
+      "loss": 0.8008,
+      "step": 3217
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0520309984634e-06,
+      "loss": 0.8044,
+      "step": 3218
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.046189985038807e-06,
+      "loss": 0.7749,
+      "step": 3219
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.040350075502324e-06,
+      "loss": 0.8457,
+      "step": 3220
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.03451127203642e-06,
+      "loss": 0.7769,
+      "step": 3221
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.028673576823155e-06,
+      "loss": 0.7233,
+      "step": 3222
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.022836992044174e-06,
+      "loss": 0.7764,
+      "step": 3223
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.017001519880704e-06,
+      "loss": 0.7644,
+      "step": 3224
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.011167162513558e-06,
+      "loss": 0.7659,
+      "step": 3225
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.005333922123135e-06,
+      "loss": 0.7771,
+      "step": 3226
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.999501800889419e-06,
+      "loss": 0.8018,
+      "step": 3227
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.993670800991961e-06,
+      "loss": 0.7842,
+      "step": 3228
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.987840924609905e-06,
+      "loss": 0.7571,
+      "step": 3229
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.982012173921984e-06,
+      "loss": 0.7896,
+      "step": 3230
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.976184551106484e-06,
+      "loss": 0.8115,
+      "step": 3231
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.970358058341294e-06,
+      "loss": 0.7834,
+      "step": 3232
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.964532697803869e-06,
+      "loss": 0.7598,
+      "step": 3233
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.958708471671245e-06,
+      "loss": 0.7693,
+      "step": 3234
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.952885382120032e-06,
+      "loss": 0.8125,
+      "step": 3235
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.947063431326418e-06,
+      "loss": 0.7856,
+      "step": 3236
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.941242621466156e-06,
+      "loss": 0.7795,
+      "step": 3237
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9354229547145855e-06,
+      "loss": 0.7583,
+      "step": 3238
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.929604433246615e-06,
+      "loss": 0.76,
+      "step": 3239
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.923787059236717e-06,
+      "loss": 0.7559,
+      "step": 3240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.917970834858944e-06,
+      "loss": 0.7469,
+      "step": 3241
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.912155762286917e-06,
+      "loss": 0.7764,
+      "step": 3242
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.90634184369383e-06,
+      "loss": 0.769,
+      "step": 3243
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.900529081252432e-06,
+      "loss": 0.8135,
+      "step": 3244
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.894717477135055e-06,
+      "loss": 0.7913,
+      "step": 3245
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.888907033513592e-06,
+      "loss": 0.7585,
+      "step": 3246
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.883097752559502e-06,
+      "loss": 0.7776,
+      "step": 3247
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.877289636443809e-06,
+      "loss": 0.7356,
+      "step": 3248
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8714826873371095e-06,
+      "loss": 0.7563,
+      "step": 3249
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8656769074095484e-06,
+      "loss": 0.7915,
+      "step": 3250
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.859872298830846e-06,
+      "loss": 0.7671,
+      "step": 3251
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.854068863770287e-06,
+      "loss": 0.7556,
+      "step": 3252
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.848266604396702e-06,
+      "loss": 0.781,
+      "step": 3253
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.842465522878496e-06,
+      "loss": 0.7646,
+      "step": 3254
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8366656213836354e-06,
+      "loss": 0.7771,
+      "step": 3255
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.830866902079633e-06,
+      "loss": 0.8103,
+      "step": 3256
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.825069367133569e-06,
+      "loss": 0.7732,
+      "step": 3257
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8192730187120805e-06,
+      "loss": 0.7393,
+      "step": 3258
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.813477858981363e-06,
+      "loss": 0.7842,
+      "step": 3259
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8076838901071555e-06,
+      "loss": 0.7585,
+      "step": 3260
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.801891114254766e-06,
+      "loss": 0.7688,
+      "step": 3261
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.796099533589055e-06,
+      "loss": 0.7693,
+      "step": 3262
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7903091502744255e-06,
+      "loss": 0.7815,
+      "step": 3263
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.784519966474845e-06,
+      "loss": 0.7737,
+      "step": 3264
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.778731984353831e-06,
+      "loss": 0.75,
+      "step": 3265
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.772945206074441e-06,
+      "loss": 0.7864,
+      "step": 3266
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7671596337992965e-06,
+      "loss": 0.7886,
+      "step": 3267
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.76137526969056e-06,
+      "loss": 0.79,
+      "step": 3268
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.755592115909946e-06,
+      "loss": 0.7737,
+      "step": 3269
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7498101746187135e-06,
+      "loss": 0.7283,
+      "step": 3270
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.744029447977678e-06,
+      "loss": 0.7551,
+      "step": 3271
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.738249938147184e-06,
+      "loss": 0.7742,
+      "step": 3272
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.732471647287133e-06,
+      "loss": 0.8347,
+      "step": 3273
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.726694577556973e-06,
+      "loss": 0.7576,
+      "step": 3274
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.720918731115691e-06,
+      "loss": 0.8025,
+      "step": 3275
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.715144110121811e-06,
+      "loss": 0.7441,
+      "step": 3276
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.70937071673341e-06,
+      "loss": 0.78,
+      "step": 3277
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7035985531081015e-06,
+      "loss": 0.7688,
+      "step": 3278
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.697827621403039e-06,
+      "loss": 0.7822,
+      "step": 3279
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.692057923774914e-06,
+      "loss": 0.7355,
+      "step": 3280
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6862894623799645e-06,
+      "loss": 0.7834,
+      "step": 3281
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.680522239373953e-06,
+      "loss": 0.7634,
+      "step": 3282
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6747562569121915e-06,
+      "loss": 0.7815,
+      "step": 3283
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.668991517149525e-06,
+      "loss": 0.7451,
+      "step": 3284
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.663228022240329e-06,
+      "loss": 0.7903,
+      "step": 3285
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.657465774338519e-06,
+      "loss": 0.7332,
+      "step": 3286
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.65170477559755e-06,
+      "loss": 0.7571,
+      "step": 3287
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.645945028170393e-06,
+      "loss": 0.7251,
+      "step": 3288
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.640186534209565e-06,
+      "loss": 0.7815,
+      "step": 3289
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.634429295867114e-06,
+      "loss": 0.8008,
+      "step": 3290
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.628673315294616e-06,
+      "loss": 0.7568,
+      "step": 3291
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.622918594643175e-06,
+      "loss": 0.8086,
+      "step": 3292
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6171651360634255e-06,
+      "loss": 0.7731,
+      "step": 3293
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.611412941705537e-06,
+      "loss": 0.7864,
+      "step": 3294
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.605662013719191e-06,
+      "loss": 0.7708,
+      "step": 3295
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5999123542536125e-06,
+      "loss": 0.7634,
+      "step": 3296
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.594163965457545e-06,
+      "loss": 0.7883,
+      "step": 3297
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.588416849479252e-06,
+      "loss": 0.7998,
+      "step": 3298
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.58267100846653e-06,
+      "loss": 0.7476,
+      "step": 3299
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.576926444566698e-06,
+      "loss": 0.7615,
+      "step": 3300
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.571183159926591e-06,
+      "loss": 0.7938,
+      "step": 3301
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.565441156692571e-06,
+      "loss": 0.7664,
+      "step": 3302
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.559700437010528e-06,
+      "loss": 0.8103,
+      "step": 3303
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5539610030258535e-06,
+      "loss": 0.782,
+      "step": 3304
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.548222856883476e-06,
+      "loss": 0.7698,
+      "step": 3305
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.542486000727835e-06,
+      "loss": 0.7539,
+      "step": 3306
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.536750436702891e-06,
+      "loss": 0.7783,
+      "step": 3307
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.531016166952118e-06,
+      "loss": 0.7727,
+      "step": 3308
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.525283193618508e-06,
+      "loss": 0.8135,
+      "step": 3309
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.519551518844576e-06,
+      "loss": 0.7527,
+      "step": 3310
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.513821144772333e-06,
+      "loss": 0.7654,
+      "step": 3311
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.508092073543322e-06,
+      "loss": 0.7776,
+      "step": 3312
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.502364307298592e-06,
+      "loss": 0.7903,
+      "step": 3313
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.496637848178704e-06,
+      "loss": 0.7395,
+      "step": 3314
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.49091269832373e-06,
+      "loss": 0.7347,
+      "step": 3315
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.485188859873262e-06,
+      "loss": 0.7686,
+      "step": 3316
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.479466334966384e-06,
+      "loss": 0.7659,
+      "step": 3317
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4737451257417015e-06,
+      "loss": 0.7799,
+      "step": 3318
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.468025234337332e-06,
+      "loss": 0.7817,
+      "step": 3319
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4623066628908844e-06,
+      "loss": 0.7876,
+      "step": 3320
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.456589413539491e-06,
+      "loss": 0.7683,
+      "step": 3321
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.450873488419782e-06,
+      "loss": 0.791,
+      "step": 3322
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.445158889667895e-06,
+      "loss": 0.8093,
+      "step": 3323
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.439445619419469e-06,
+      "loss": 0.7544,
+      "step": 3324
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.433733679809648e-06,
+      "loss": 0.8188,
+      "step": 3325
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.428023072973087e-06,
+      "loss": 0.7876,
+      "step": 3326
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.422313801043922e-06,
+      "loss": 0.804,
+      "step": 3327
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4166058661558104e-06,
+      "loss": 0.803,
+      "step": 3328
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.410899270441905e-06,
+      "loss": 0.8184,
+      "step": 3329
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.40519401603485e-06,
+      "loss": 0.7693,
+      "step": 3330
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.399490105066799e-06,
+      "loss": 0.7542,
+      "step": 3331
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3937875396693994e-06,
+      "loss": 0.762,
+      "step": 3332
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.38808632197379e-06,
+      "loss": 0.7856,
+      "step": 3333
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.382386454110613e-06,
+      "loss": 0.7803,
+      "step": 3334
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.376687938210007e-06,
+      "loss": 0.8074,
+      "step": 3335
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.370990776401598e-06,
+      "loss": 0.8088,
+      "step": 3336
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.365294970814513e-06,
+      "loss": 0.7432,
+      "step": 3337
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.359600523577369e-06,
+      "loss": 0.8008,
+      "step": 3338
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.353907436818283e-06,
+      "loss": 0.782,
+      "step": 3339
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.348215712664844e-06,
+      "loss": 0.8047,
+      "step": 3340
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.342525353244151e-06,
+      "loss": 0.752,
+      "step": 3341
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.336836360682789e-06,
+      "loss": 0.7922,
+      "step": 3342
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.331148737106822e-06,
+      "loss": 0.7686,
+      "step": 3343
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.325462484641815e-06,
+      "loss": 0.7749,
+      "step": 3344
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.319777605412816e-06,
+      "loss": 0.7097,
+      "step": 3345
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.314094101544356e-06,
+      "loss": 0.7781,
+      "step": 3346
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.308411975160456e-06,
+      "loss": 0.8169,
+      "step": 3347
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.302731228384628e-06,
+      "loss": 0.7625,
+      "step": 3348
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2970518633398514e-06,
+      "loss": 0.7798,
+      "step": 3349
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2913738821486055e-06,
+      "loss": 0.7588,
+      "step": 3350
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.285697286932847e-06,
+      "loss": 0.8472,
+      "step": 3351
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.280022079814012e-06,
+      "loss": 0.7788,
+      "step": 3352
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.27434826291302e-06,
+      "loss": 0.7827,
+      "step": 3353
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.268675838350273e-06,
+      "loss": 0.7485,
+      "step": 3354
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2630048082456545e-06,
+      "loss": 0.75,
+      "step": 3355
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.257335174718515e-06,
+      "loss": 0.7402,
+      "step": 3356
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.251666939887695e-06,
+      "loss": 0.7615,
+      "step": 3357
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2460001058715095e-06,
+      "loss": 0.7446,
+      "step": 3358
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.240334674787746e-06,
+      "loss": 0.7542,
+      "step": 3359
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.234670648753672e-06,
+      "loss": 0.7649,
+      "step": 3360
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.229008029886035e-06,
+      "loss": 0.7773,
+      "step": 3361
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.223346820301038e-06,
+      "loss": 0.7603,
+      "step": 3362
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.217687022114377e-06,
+      "loss": 0.7666,
+      "step": 3363
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2120286374412165e-06,
+      "loss": 0.8184,
+      "step": 3364
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.206371668396179e-06,
+      "loss": 0.7568,
+      "step": 3365
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.200716117093375e-06,
+      "loss": 0.7507,
+      "step": 3366
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.19506198564638e-06,
+      "loss": 0.7634,
+      "step": 3367
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.189409276168232e-06,
+      "loss": 0.7859,
+      "step": 3368
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.183757990771445e-06,
+      "loss": 0.7651,
+      "step": 3369
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.178108131567999e-06,
+      "loss": 0.8145,
+      "step": 3370
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.172459700669346e-06,
+      "loss": 0.762,
+      "step": 3371
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1668127001863895e-06,
+      "loss": 0.7727,
+      "step": 3372
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.161167132229513e-06,
+      "loss": 0.7944,
+      "step": 3373
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.155522998908559e-06,
+      "loss": 0.8076,
+      "step": 3374
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.149880302332833e-06,
+      "loss": 0.7463,
+      "step": 3375
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1442390446111065e-06,
+      "loss": 0.7969,
+      "step": 3376
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1385992278516135e-06,
+      "loss": 0.7534,
+      "step": 3377
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1329608541620426e-06,
+      "loss": 0.8125,
+      "step": 3378
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.127323925649551e-06,
+      "loss": 0.7805,
+      "step": 3379
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.121688444420754e-06,
+      "loss": 0.7944,
+      "step": 3380
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.116054412581721e-06,
+      "loss": 0.7729,
+      "step": 3381
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.110421832237987e-06,
+      "loss": 0.7935,
+      "step": 3382
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.104790705494546e-06,
+      "loss": 0.7563,
+      "step": 3383
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.099161034455834e-06,
+      "loss": 0.8096,
+      "step": 3384
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0935328212257586e-06,
+      "loss": 0.7627,
+      "step": 3385
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.087906067907676e-06,
+      "loss": 0.7778,
+      "step": 3386
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0822807766044e-06,
+      "loss": 0.7654,
+      "step": 3387
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.076656949418192e-06,
+      "loss": 0.7554,
+      "step": 3388
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.071034588450773e-06,
+      "loss": 0.7781,
+      "step": 3389
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.065413695803316e-06,
+      "loss": 0.7617,
+      "step": 3390
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.059794273576434e-06,
+      "loss": 0.7351,
+      "step": 3391
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0541763238702046e-06,
+      "loss": 0.7808,
+      "step": 3392
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.048559848784153e-06,
+      "loss": 0.7771,
+      "step": 3393
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.042944850417238e-06,
+      "loss": 0.7786,
+      "step": 3394
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.037331330867888e-06,
+      "loss": 0.7959,
+      "step": 3395
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.031719292233968e-06,
+      "loss": 0.7637,
+      "step": 3396
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.026108736612786e-06,
+      "loss": 0.8188,
+      "step": 3397
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.020499666101103e-06,
+      "loss": 0.7683,
+      "step": 3398
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.014892082795127e-06,
+      "loss": 0.761,
+      "step": 3399
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.009285988790497e-06,
+      "loss": 0.7659,
+      "step": 3400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0036813861823076e-06,
+      "loss": 0.7493,
+      "step": 3401
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.998078277065091e-06,
+      "loss": 0.7925,
+      "step": 3402
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.992476663532828e-06,
+      "loss": 0.7339,
+      "step": 3403
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.986876547678928e-06,
+      "loss": 0.7556,
+      "step": 3404
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.981277931596252e-06,
+      "loss": 0.7722,
+      "step": 3405
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.975680817377099e-06,
+      "loss": 0.75,
+      "step": 3406
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.970085207113197e-06,
+      "loss": 0.7537,
+      "step": 3407
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.96449110289572e-06,
+      "loss": 0.7874,
+      "step": 3408
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.958898506815283e-06,
+      "loss": 0.7847,
+      "step": 3409
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.953307420961929e-06,
+      "loss": 0.7732,
+      "step": 3410
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.947717847425141e-06,
+      "loss": 0.8083,
+      "step": 3411
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.942129788293838e-06,
+      "loss": 0.7576,
+      "step": 3412
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9365432456563666e-06,
+      "loss": 0.7432,
+      "step": 3413
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.930958221600514e-06,
+      "loss": 0.7678,
+      "step": 3414
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.925374718213498e-06,
+      "loss": 0.7903,
+      "step": 3415
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.919792737581962e-06,
+      "loss": 0.7383,
+      "step": 3416
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9142122817919885e-06,
+      "loss": 0.7893,
+      "step": 3417
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.908633352929087e-06,
+      "loss": 0.7395,
+      "step": 3418
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9030559530782e-06,
+      "loss": 0.7473,
+      "step": 3419
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.897480084323688e-06,
+      "loss": 0.7888,
+      "step": 3420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.891905748749349e-06,
+      "loss": 0.7747,
+      "step": 3421
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.886332948438411e-06,
+      "loss": 0.7366,
+      "step": 3422
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.880761685473513e-06,
+      "loss": 0.7422,
+      "step": 3423
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.875191961936733e-06,
+      "loss": 0.7808,
+      "step": 3424
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.869623779909572e-06,
+      "loss": 0.7446,
+      "step": 3425
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.864057141472948e-06,
+      "loss": 0.7654,
+      "step": 3426
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8584920487072105e-06,
+      "loss": 0.7732,
+      "step": 3427
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.85292850369213e-06,
+      "loss": 0.8,
+      "step": 3428
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8473665085068876e-06,
+      "loss": 0.7659,
+      "step": 3429
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.841806065230099e-06,
+      "loss": 0.7881,
+      "step": 3430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.836247175939796e-06,
+      "loss": 0.7297,
+      "step": 3431
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.830689842713424e-06,
+      "loss": 0.7446,
+      "step": 3432
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.825134067627856e-06,
+      "loss": 0.7891,
+      "step": 3433
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.819579852759374e-06,
+      "loss": 0.7991,
+      "step": 3434
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.814027200183685e-06,
+      "loss": 0.8125,
+      "step": 3435
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.808476111975904e-06,
+      "loss": 0.8064,
+      "step": 3436
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.802926590210565e-06,
+      "loss": 0.77,
+      "step": 3437
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7973786369616236e-06,
+      "loss": 0.7837,
+      "step": 3438
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.791832254302433e-06,
+      "loss": 0.7869,
+      "step": 3439
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7862874443057736e-06,
+      "loss": 0.7424,
+      "step": 3440
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.780744209043838e-06,
+      "loss": 0.8167,
+      "step": 3441
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.775202550588217e-06,
+      "loss": 0.7515,
+      "step": 3442
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.769662471009925e-06,
+      "loss": 0.7393,
+      "step": 3443
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7641239723793815e-06,
+      "loss": 0.7837,
+      "step": 3444
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.758587056766417e-06,
+      "loss": 0.7573,
+      "step": 3445
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.753051726240268e-06,
+      "loss": 0.7695,
+      "step": 3446
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.747517982869587e-06,
+      "loss": 0.7554,
+      "step": 3447
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.741985828722415e-06,
+      "loss": 0.8262,
+      "step": 3448
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.736455265866216e-06,
+      "loss": 0.7454,
+      "step": 3449
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.730926296367852e-06,
+      "loss": 0.7678,
+      "step": 3450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.725398922293597e-06,
+      "loss": 0.7495,
+      "step": 3451
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7198731457091135e-06,
+      "loss": 0.7998,
+      "step": 3452
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7143489686794825e-06,
+      "loss": 0.7812,
+      "step": 3453
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7088263932691835e-06,
+      "loss": 0.7273,
+      "step": 3454
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.703305421542087e-06,
+      "loss": 0.7898,
+      "step": 3455
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.697786055561478e-06,
+      "loss": 0.7859,
+      "step": 3456
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.692268297390037e-06,
+      "loss": 0.8005,
+      "step": 3457
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.686752149089836e-06,
+      "loss": 0.7737,
+      "step": 3458
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.681237612722357e-06,
+      "loss": 0.7522,
+      "step": 3459
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.675724690348471e-06,
+      "loss": 0.7507,
+      "step": 3460
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.670213384028452e-06,
+      "loss": 0.7605,
+      "step": 3461
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.664703695821965e-06,
+      "loss": 0.7395,
+      "step": 3462
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.659195627788079e-06,
+      "loss": 0.7505,
+      "step": 3463
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.65368918198524e-06,
+      "loss": 0.7922,
+      "step": 3464
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.648184360471305e-06,
+      "loss": 0.7656,
+      "step": 3465
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.642681165303516e-06,
+      "loss": 0.8008,
+      "step": 3466
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6371795985385155e-06,
+      "loss": 0.7466,
+      "step": 3467
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.63167966223232e-06,
+      "loss": 0.8174,
+      "step": 3468
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.626181358440352e-06,
+      "loss": 0.7505,
+      "step": 3469
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.620684689217426e-06,
+      "loss": 0.751,
+      "step": 3470
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.615189656617727e-06,
+      "loss": 0.77,
+      "step": 3471
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.609696262694849e-06,
+      "loss": 0.7661,
+      "step": 3472
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.604204509501761e-06,
+      "loss": 0.7644,
+      "step": 3473
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.598714399090825e-06,
+      "loss": 0.7214,
+      "step": 3474
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5932259335137865e-06,
+      "loss": 0.8059,
+      "step": 3475
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.587739114821779e-06,
+      "loss": 0.7795,
+      "step": 3476
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.582253945065312e-06,
+      "loss": 0.7859,
+      "step": 3477
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.576770426294289e-06,
+      "loss": 0.7615,
+      "step": 3478
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.571288560557997e-06,
+      "loss": 0.7939,
+      "step": 3479
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.565808349905091e-06,
+      "loss": 0.8091,
+      "step": 3480
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.560329796383622e-06,
+      "loss": 0.7803,
+      "step": 3481
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.554852902041017e-06,
+      "loss": 0.7305,
+      "step": 3482
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.549377668924087e-06,
+      "loss": 0.7705,
+      "step": 3483
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.543904099079011e-06,
+      "loss": 0.7754,
+      "step": 3484
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.538432194551356e-06,
+      "loss": 0.7495,
+      "step": 3485
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.532961957386067e-06,
+      "loss": 0.7487,
+      "step": 3486
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.527493389627457e-06,
+      "loss": 0.7651,
+      "step": 3487
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.522026493319225e-06,
+      "loss": 0.7417,
+      "step": 3488
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.516561270504439e-06,
+      "loss": 0.7729,
+      "step": 3489
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.511097723225547e-06,
+      "loss": 0.7606,
+      "step": 3490
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.505635853524367e-06,
+      "loss": 0.7639,
+      "step": 3491
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.500175663442093e-06,
+      "loss": 0.7334,
+      "step": 3492
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.494717155019285e-06,
+      "loss": 0.7905,
+      "step": 3493
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.489260330295878e-06,
+      "loss": 0.8013,
+      "step": 3494
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.483805191311187e-06,
+      "loss": 0.8025,
+      "step": 3495
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.478351740103876e-06,
+      "loss": 0.7551,
+      "step": 3496
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.472899978711998e-06,
+      "loss": 0.7478,
+      "step": 3497
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.467449909172966e-06,
+      "loss": 0.783,
+      "step": 3498
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.462001533523566e-06,
+      "loss": 0.8,
+      "step": 3499
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.456554853799941e-06,
+      "loss": 0.772,
+      "step": 3500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.451109872037608e-06,
+      "loss": 0.7341,
+      "step": 3501
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.44566659027145e-06,
+      "loss": 0.7834,
+      "step": 3502
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.440225010535707e-06,
+      "loss": 0.7437,
+      "step": 3503
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.434785134863991e-06,
+      "loss": 0.7622,
+      "step": 3504
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.429346965289276e-06,
+      "loss": 0.7954,
+      "step": 3505
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.423910503843894e-06,
+      "loss": 0.7947,
+      "step": 3506
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.418475752559544e-06,
+      "loss": 0.8206,
+      "step": 3507
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.413042713467286e-06,
+      "loss": 0.7458,
+      "step": 3508
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.407611388597529e-06,
+      "loss": 0.7812,
+      "step": 3509
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.402181779980053e-06,
+      "loss": 0.7488,
+      "step": 3510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.396753889644002e-06,
+      "loss": 0.7751,
+      "step": 3511
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3913277196178544e-06,
+      "loss": 0.7449,
+      "step": 3512
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3859032719294705e-06,
+      "loss": 0.7415,
+      "step": 3513
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.380480548606053e-06,
+      "loss": 0.7617,
+      "step": 3514
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.375059551674169e-06,
+      "loss": 0.7537,
+      "step": 3515
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.369640283159729e-06,
+      "loss": 0.814,
+      "step": 3516
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.364222745088008e-06,
+      "loss": 0.7759,
+      "step": 3517
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.358806939483628e-06,
+      "loss": 0.8015,
+      "step": 3518
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.353392868370569e-06,
+      "loss": 0.7476,
+      "step": 3519
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.347980533772157e-06,
+      "loss": 0.7461,
+      "step": 3520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.342569937711077e-06,
+      "loss": 0.7573,
+      "step": 3521
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.337161082209352e-06,
+      "loss": 0.748,
+      "step": 3522
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3317539692883645e-06,
+      "loss": 0.7517,
+      "step": 3523
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.326348600968847e-06,
+      "loss": 0.7546,
+      "step": 3524
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3209449792708675e-06,
+      "loss": 0.7686,
+      "step": 3525
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.315543106213856e-06,
+      "loss": 0.7527,
+      "step": 3526
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.310142983816583e-06,
+      "loss": 0.7517,
+      "step": 3527
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3047446140971586e-06,
+      "loss": 0.7573,
+      "step": 3528
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.299347999073047e-06,
+      "loss": 0.7683,
+      "step": 3529
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2939531407610535e-06,
+      "loss": 0.7532,
+      "step": 3530
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2885600411773304e-06,
+      "loss": 0.7852,
+      "step": 3531
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.283168702337362e-06,
+      "loss": 0.7798,
+      "step": 3532
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.277779126255984e-06,
+      "loss": 0.802,
+      "step": 3533
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2723913149473735e-06,
+      "loss": 0.5719,
+      "step": 3534
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.267005270425043e-06,
+      "loss": 0.6163,
+      "step": 3535
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.261620994701848e-06,
+      "loss": 0.5488,
+      "step": 3536
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.256238489789987e-06,
+      "loss": 0.5618,
+      "step": 3537
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.250857757700982e-06,
+      "loss": 0.5393,
+      "step": 3538
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.245478800445709e-06,
+      "loss": 0.5897,
+      "step": 3539
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.240101620034377e-06,
+      "loss": 0.533,
+      "step": 3540
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.234726218476519e-06,
+      "loss": 0.5457,
+      "step": 3541
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.229352597781018e-06,
+      "loss": 0.5514,
+      "step": 3542
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.223980759956089e-06,
+      "loss": 0.564,
+      "step": 3543
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2186107070092685e-06,
+      "loss": 0.5192,
+      "step": 3544
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2132424409474395e-06,
+      "loss": 0.5493,
+      "step": 3545
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.207875963776814e-06,
+      "loss": 0.5183,
+      "step": 3546
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2025112775029376e-06,
+      "loss": 0.4951,
+      "step": 3547
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.197148384130673e-06,
+      "loss": 0.5535,
+      "step": 3548
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.191787285664229e-06,
+      "loss": 0.5326,
+      "step": 3549
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1864279841071385e-06,
+      "loss": 0.5481,
+      "step": 3550
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.18107048146226e-06,
+      "loss": 0.5281,
+      "step": 3551
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.175714779731782e-06,
+      "loss": 0.552,
+      "step": 3552
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.170360880917223e-06,
+      "loss": 0.491,
+      "step": 3553
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1650087870194185e-06,
+      "loss": 0.5151,
+      "step": 3554
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.159658500038538e-06,
+      "loss": 0.5219,
+      "step": 3555
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.154310021974076e-06,
+      "loss": 0.5358,
+      "step": 3556
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.148963354824843e-06,
+      "loss": 0.4895,
+      "step": 3557
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.143618500588978e-06,
+      "loss": 0.5929,
+      "step": 3558
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.138275461263949e-06,
+      "loss": 0.4962,
+      "step": 3559
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.132934238846532e-06,
+      "loss": 0.5111,
+      "step": 3560
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.127594835332831e-06,
+      "loss": 0.5265,
+      "step": 3561
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1222572527182725e-06,
+      "loss": 0.5566,
+      "step": 3562
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1169214929976e-06,
+      "loss": 0.5621,
+      "step": 3563
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.111587558164877e-06,
+      "loss": 0.5303,
+      "step": 3564
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.106255450213483e-06,
+      "loss": 0.5508,
+      "step": 3565
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1009251711361196e-06,
+      "loss": 0.578,
+      "step": 3566
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.095596722924793e-06,
+      "loss": 0.5393,
+      "step": 3567
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.090270107570837e-06,
+      "loss": 0.5531,
+      "step": 3568
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.084945327064902e-06,
+      "loss": 0.5342,
+      "step": 3569
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.079622383396938e-06,
+      "loss": 0.562,
+      "step": 3570
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.074301278556223e-06,
+      "loss": 0.5166,
+      "step": 3571
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.068982014531345e-06,
+      "loss": 0.5591,
+      "step": 3572
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0636645933101954e-06,
+      "loss": 0.5502,
+      "step": 3573
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.058349016879989e-06,
+      "loss": 0.5796,
+      "step": 3574
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.053035287227244e-06,
+      "loss": 0.505,
+      "step": 3575
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.047723406337787e-06,
+      "loss": 0.5494,
+      "step": 3576
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.042413376196758e-06,
+      "loss": 0.4982,
+      "step": 3577
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.037105198788605e-06,
+      "loss": 0.525,
+      "step": 3578
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0317988760970806e-06,
+      "loss": 0.4946,
+      "step": 3579
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.02649441010525e-06,
+      "loss": 0.5676,
+      "step": 3580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.021191802795476e-06,
+      "loss": 0.525,
+      "step": 3581
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.015891056149438e-06,
+      "loss": 0.5363,
+      "step": 3582
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.010592172148105e-06,
+      "loss": 0.5112,
+      "step": 3583
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.005295152771765e-06,
+      "loss": 0.5427,
+      "step": 3584
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.517,
+      "step": 3585
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.994706715811698e-06,
+      "loss": 0.531,
+      "step": 3586
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.989415302185045e-06,
+      "loss": 0.5195,
+      "step": 3587
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.984125761097537e-06,
+      "loss": 0.5284,
+      "step": 3588
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.978838094525954e-06,
+      "loss": 0.5415,
+      "step": 3589
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9735523044463905e-06,
+      "loss": 0.4941,
+      "step": 3590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.968268392834236e-06,
+      "loss": 0.5266,
+      "step": 3591
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9629863616641735e-06,
+      "loss": 0.5331,
+      "step": 3592
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.957706212910189e-06,
+      "loss": 0.5909,
+      "step": 3593
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.95242794854556e-06,
+      "loss": 0.5135,
+      "step": 3594
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.947151570542866e-06,
+      "loss": 0.5713,
+      "step": 3595
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.941877080873973e-06,
+      "loss": 0.5319,
+      "step": 3596
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.936604481510046e-06,
+      "loss": 0.5291,
+      "step": 3597
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9313337744215476e-06,
+      "loss": 0.5779,
+      "step": 3598
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.926064961578223e-06,
+      "loss": 0.5344,
+      "step": 3599
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.920798044949119e-06,
+      "loss": 0.5209,
+      "step": 3600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.915533026502573e-06,
+      "loss": 0.5197,
+      "step": 3601
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.910269908206204e-06,
+      "loss": 0.5353,
+      "step": 3602
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.90500869202693e-06,
+      "loss": 0.5146,
+      "step": 3603
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.899749379930959e-06,
+      "loss": 0.5398,
+      "step": 3604
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.894491973883778e-06,
+      "loss": 0.5249,
+      "step": 3605
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.889236475850169e-06,
+      "loss": 0.5315,
+      "step": 3606
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8839828877942e-06,
+      "loss": 0.5787,
+      "step": 3607
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.878731211679224e-06,
+      "loss": 0.5199,
+      "step": 3608
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8734814494678815e-06,
+      "loss": 0.5555,
+      "step": 3609
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.868233603122094e-06,
+      "loss": 0.4938,
+      "step": 3610
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.862987674603077e-06,
+      "loss": 0.5009,
+      "step": 3611
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.857743665871309e-06,
+      "loss": 0.5278,
+      "step": 3612
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.852501578886572e-06,
+      "loss": 0.5597,
+      "step": 3613
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.847261415607922e-06,
+      "loss": 0.5464,
+      "step": 3614
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.842023177993689e-06,
+      "loss": 0.5413,
+      "step": 3615
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.836786868001494e-06,
+      "loss": 0.5166,
+      "step": 3616
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.831552487588236e-06,
+      "loss": 0.5333,
+      "step": 3617
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.826320038710084e-06,
+      "loss": 0.5765,
+      "step": 3618
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.821089523322496e-06,
+      "loss": 0.5128,
+      "step": 3619
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.815860943380204e-06,
+      "loss": 0.5477,
+      "step": 3620
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.81063430083721e-06,
+      "loss": 0.5543,
+      "step": 3621
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.805409597646801e-06,
+      "loss": 0.5604,
+      "step": 3622
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.800186835761535e-06,
+      "loss": 0.5627,
+      "step": 3623
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.794966017133248e-06,
+      "loss": 0.5248,
+      "step": 3624
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.789747143713044e-06,
+      "loss": 0.5233,
+      "step": 3625
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7845302174513046e-06,
+      "loss": 0.5121,
+      "step": 3626
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.779315240297685e-06,
+      "loss": 0.5302,
+      "step": 3627
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.774102214201103e-06,
+      "loss": 0.4915,
+      "step": 3628
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.768891141109756e-06,
+      "loss": 0.5223,
+      "step": 3629
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.763682022971113e-06,
+      "loss": 0.5076,
+      "step": 3630
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.758474861731902e-06,
+      "loss": 0.5903,
+      "step": 3631
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.753269659338129e-06,
+      "loss": 0.5251,
+      "step": 3632
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.748066417735068e-06,
+      "loss": 0.5195,
+      "step": 3633
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.742865138867251e-06,
+      "loss": 0.5449,
+      "step": 3634
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.737665824678485e-06,
+      "loss": 0.5358,
+      "step": 3635
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.732468477111844e-06,
+      "loss": 0.5172,
+      "step": 3636
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7272730981096595e-06,
+      "loss": 0.5166,
+      "step": 3637
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.722079689613533e-06,
+      "loss": 0.5293,
+      "step": 3638
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.716888253564332e-06,
+      "loss": 0.5129,
+      "step": 3639
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.711698791902174e-06,
+      "loss": 0.5394,
+      "step": 3640
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.706511306566452e-06,
+      "loss": 0.4974,
+      "step": 3641
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.701325799495817e-06,
+      "loss": 0.5155,
+      "step": 3642
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6961422726281815e-06,
+      "loss": 0.5507,
+      "step": 3643
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6909607279007094e-06,
+      "loss": 0.4968,
+      "step": 3644
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.685781167249836e-06,
+      "loss": 0.5333,
+      "step": 3645
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6806035926112495e-06,
+      "loss": 0.4982,
+      "step": 3646
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.67542800591989e-06,
+      "loss": 0.5549,
+      "step": 3647
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.670254409109965e-06,
+      "loss": 0.5201,
+      "step": 3648
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6650828041149385e-06,
+      "loss": 0.564,
+      "step": 3649
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.659913192867516e-06,
+      "loss": 0.5664,
+      "step": 3650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6547455772996745e-06,
+      "loss": 0.5299,
+      "step": 3651
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6495799593426335e-06,
+      "loss": 0.537,
+      "step": 3652
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.644416340926875e-06,
+      "loss": 0.4938,
+      "step": 3653
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.639254723982127e-06,
+      "loss": 0.5483,
+      "step": 3654
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.634095110437377e-06,
+      "loss": 0.5477,
+      "step": 3655
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.628937502220849e-06,
+      "loss": 0.5148,
+      "step": 3656
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.623781901260034e-06,
+      "loss": 0.5116,
+      "step": 3657
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.618628309481664e-06,
+      "loss": 0.5138,
+      "step": 3658
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.613476728811728e-06,
+      "loss": 0.5264,
+      "step": 3659
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.608327161175448e-06,
+      "loss": 0.5096,
+      "step": 3660
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6031796084973066e-06,
+      "loss": 0.5114,
+      "step": 3661
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.598034072701037e-06,
+      "loss": 0.5242,
+      "step": 3662
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.592890555709603e-06,
+      "loss": 0.5203,
+      "step": 3663
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.587749059445225e-06,
+      "loss": 0.5322,
+      "step": 3664
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.582609585829372e-06,
+      "loss": 0.5509,
+      "step": 3665
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.577472136782741e-06,
+      "loss": 0.5317,
+      "step": 3666
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.57233671422529e-06,
+      "loss": 0.5072,
+      "step": 3667
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.567203320076208e-06,
+      "loss": 0.5262,
+      "step": 3668
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.562071956253931e-06,
+      "loss": 0.5803,
+      "step": 3669
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.556942624676138e-06,
+      "loss": 0.5339,
+      "step": 3670
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.551815327259749e-06,
+      "loss": 0.5475,
+      "step": 3671
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.54669006592091e-06,
+      "loss": 0.5554,
+      "step": 3672
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.541566842575022e-06,
+      "loss": 0.5474,
+      "step": 3673
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.536445659136721e-06,
+      "loss": 0.5464,
+      "step": 3674
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5313265175198805e-06,
+      "loss": 0.549,
+      "step": 3675
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.526209419637602e-06,
+      "loss": 0.5171,
+      "step": 3676
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.521094367402233e-06,
+      "loss": 0.5679,
+      "step": 3677
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.515981362725361e-06,
+      "loss": 0.5453,
+      "step": 3678
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5108704075177914e-06,
+      "loss": 0.5562,
+      "step": 3679
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.505761503689576e-06,
+      "loss": 0.536,
+      "step": 3680
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.500654653150001e-06,
+      "loss": 0.5438,
+      "step": 3681
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.4955498578075785e-06,
+      "loss": 0.5385,
+      "step": 3682
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.490447119570056e-06,
+      "loss": 0.5698,
+      "step": 3683
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.485346440344418e-06,
+      "loss": 0.5177,
+      "step": 3684
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.480247822036864e-06,
+      "loss": 0.5461,
+      "step": 3685
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.475151266552836e-06,
+      "loss": 0.5565,
+      "step": 3686
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.470056775797008e-06,
+      "loss": 0.5258,
+      "step": 3687
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.464964351673268e-06,
+      "loss": 0.5316,
+      "step": 3688
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.459873996084742e-06,
+      "loss": 0.5109,
+      "step": 3689
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.454785710933781e-06,
+      "loss": 0.5125,
+      "step": 3690
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.449699498121967e-06,
+      "loss": 0.5317,
+      "step": 3691
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.444615359550096e-06,
+      "loss": 0.5098,
+      "step": 3692
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.439533297118198e-06,
+      "loss": 0.5134,
+      "step": 3693
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.434453312725528e-06,
+      "loss": 0.5115,
+      "step": 3694
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.429375408270554e-06,
+      "loss": 0.5144,
+      "step": 3695
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.424299585650978e-06,
+      "loss": 0.5161,
+      "step": 3696
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.419225846763717e-06,
+      "loss": 0.4713,
+      "step": 3697
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.414154193504915e-06,
+      "loss": 0.5247,
+      "step": 3698
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.409084627769932e-06,
+      "loss": 0.5493,
+      "step": 3699
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.404017151453352e-06,
+      "loss": 0.5503,
+      "step": 3700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.398951766448973e-06,
+      "loss": 0.5183,
+      "step": 3701
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.39388847464981e-06,
+      "loss": 0.5551,
+      "step": 3702
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.38882727794811e-06,
+      "loss": 0.5659,
+      "step": 3703
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.383768178235316e-06,
+      "loss": 0.5078,
+      "step": 3704
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.378711177402102e-06,
+      "loss": 0.5359,
+      "step": 3705
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.373656277338354e-06,
+      "loss": 0.5238,
+      "step": 3706
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.368603479933177e-06,
+      "loss": 0.5188,
+      "step": 3707
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.363552787074878e-06,
+      "loss": 0.5426,
+      "step": 3708
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.358504200650987e-06,
+      "loss": 0.5349,
+      "step": 3709
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.353457722548251e-06,
+      "loss": 0.5033,
+      "step": 3710
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.348413354652615e-06,
+      "loss": 0.527,
+      "step": 3711
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.343371098849247e-06,
+      "loss": 0.5295,
+      "step": 3712
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.338330957022523e-06,
+      "loss": 0.5403,
+      "step": 3713
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3332929310560265e-06,
+      "loss": 0.5353,
+      "step": 3714
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.328257022832553e-06,
+      "loss": 0.5497,
+      "step": 3715
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3232232342341095e-06,
+      "loss": 0.5704,
+      "step": 3716
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.318191567141898e-06,
+      "loss": 0.5598,
+      "step": 3717
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.313162023436343e-06,
+      "loss": 0.5417,
+      "step": 3718
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.308134604997069e-06,
+      "loss": 0.5236,
+      "step": 3719
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3031093137029e-06,
+      "loss": 0.5323,
+      "step": 3720
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.298086151431877e-06,
+      "loss": 0.5353,
+      "step": 3721
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.293065120061237e-06,
+      "loss": 0.4972,
+      "step": 3722
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.288046221467429e-06,
+      "loss": 0.575,
+      "step": 3723
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.28302945752609e-06,
+      "loss": 0.511,
+      "step": 3724
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.278014830112074e-06,
+      "loss": 0.5542,
+      "step": 3725
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.273002341099429e-06,
+      "loss": 0.5156,
+      "step": 3726
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.267991992361406e-06,
+      "loss": 0.5436,
+      "step": 3727
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.262983785770458e-06,
+      "loss": 0.5538,
+      "step": 3728
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.257977723198237e-06,
+      "loss": 0.4982,
+      "step": 3729
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.252973806515585e-06,
+      "loss": 0.5199,
+      "step": 3730
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2479720375925545e-06,
+      "loss": 0.505,
+      "step": 3731
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.242972418298393e-06,
+      "loss": 0.5095,
+      "step": 3732
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.237974950501533e-06,
+      "loss": 0.5015,
+      "step": 3733
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.232979636069616e-06,
+      "loss": 0.5049,
+      "step": 3734
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.227986476869479e-06,
+      "loss": 0.5168,
+      "step": 3735
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.222995474767141e-06,
+      "loss": 0.5554,
+      "step": 3736
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.218006631627827e-06,
+      "loss": 0.5538,
+      "step": 3737
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.213019949315949e-06,
+      "loss": 0.5015,
+      "step": 3738
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.208035429695118e-06,
+      "loss": 0.5095,
+      "step": 3739
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.203053074628126e-06,
+      "loss": 0.5369,
+      "step": 3740
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1980728859769656e-06,
+      "loss": 0.5394,
+      "step": 3741
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.193094865602816e-06,
+      "loss": 0.4861,
+      "step": 3742
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.188119015366047e-06,
+      "loss": 0.5353,
+      "step": 3743
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.183145337126216e-06,
+      "loss": 0.5273,
+      "step": 3744
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.178173832742073e-06,
+      "loss": 0.5212,
+      "step": 3745
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.173204504071545e-06,
+      "loss": 0.5056,
+      "step": 3746
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.168237352971759e-06,
+      "loss": 0.5393,
+      "step": 3747
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.163272381299024e-06,
+      "loss": 0.5585,
+      "step": 3748
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.158309590908825e-06,
+      "loss": 0.5334,
+      "step": 3749
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.153348983655844e-06,
+      "loss": 0.5082,
+      "step": 3750
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.148390561393946e-06,
+      "loss": 0.5461,
+      "step": 3751
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.14343432597617e-06,
+      "loss": 0.5168,
+      "step": 3752
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.138480279254746e-06,
+      "loss": 0.5377,
+      "step": 3753
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1335284230810845e-06,
+      "loss": 0.5073,
+      "step": 3754
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.128578759305782e-06,
+      "loss": 0.5099,
+      "step": 3755
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.123631289778601e-06,
+      "loss": 0.5142,
+      "step": 3756
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1186860163484975e-06,
+      "loss": 0.5409,
+      "step": 3757
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.113742940863604e-06,
+      "loss": 0.5262,
+      "step": 3758
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.108802065171229e-06,
+      "loss": 0.5548,
+      "step": 3759
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.103863391117862e-06,
+      "loss": 0.5331,
+      "step": 3760
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.098926920549168e-06,
+      "loss": 0.5404,
+      "step": 3761
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0939926553099855e-06,
+      "loss": 0.5679,
+      "step": 3762
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.089060597244332e-06,
+      "loss": 0.5182,
+      "step": 3763
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.084130748195405e-06,
+      "loss": 0.52,
+      "step": 3764
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.079203110005565e-06,
+      "loss": 0.5471,
+      "step": 3765
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.074277684516355e-06,
+      "loss": 0.5509,
+      "step": 3766
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.069354473568492e-06,
+      "loss": 0.5282,
+      "step": 3767
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.064433479001857e-06,
+      "loss": 0.5464,
+      "step": 3768
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0595147026555095e-06,
+      "loss": 0.506,
+      "step": 3769
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0545981463676775e-06,
+      "loss": 0.5101,
+      "step": 3770
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0496838119757635e-06,
+      "loss": 0.512,
+      "step": 3771
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0447717013163325e-06,
+      "loss": 0.5223,
+      "step": 3772
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.039861816225125e-06,
+      "loss": 0.5507,
+      "step": 3773
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034954158537049e-06,
+      "loss": 0.5161,
+      "step": 3774
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.030048730086171e-06,
+      "loss": 0.4934,
+      "step": 3775
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.025145532705738e-06,
+      "loss": 0.5493,
+      "step": 3776
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.020244568228155e-06,
+      "loss": 0.5364,
+      "step": 3777
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0153458384849915e-06,
+      "loss": 0.5454,
+      "step": 3778
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.010449345306987e-06,
+      "loss": 0.5234,
+      "step": 3779
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0055550905240455e-06,
+      "loss": 0.5104,
+      "step": 3780
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.000663075965225e-06,
+      "loss": 0.5509,
+      "step": 3781
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.995773303458757e-06,
+      "loss": 0.536,
+      "step": 3782
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9908857748320325e-06,
+      "loss": 0.5288,
+      "step": 3783
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.986000491911599e-06,
+      "loss": 0.522,
+      "step": 3784
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.98111745652317e-06,
+      "loss": 0.5243,
+      "step": 3785
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.976236670491617e-06,
+      "loss": 0.4956,
+      "step": 3786
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.971358135640971e-06,
+      "loss": 0.5253,
+      "step": 3787
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.966481853794423e-06,
+      "loss": 0.55,
+      "step": 3788
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.96160782677432e-06,
+      "loss": 0.5468,
+      "step": 3789
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.956736056402171e-06,
+      "loss": 0.5299,
+      "step": 3790
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.951866544498631e-06,
+      "loss": 0.5791,
+      "step": 3791
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.946999292883523e-06,
+      "loss": 0.5208,
+      "step": 3792
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.94213430337582e-06,
+      "loss": 0.484,
+      "step": 3793
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.937271577793646e-06,
+      "loss": 0.5963,
+      "step": 3794
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932411117954285e-06,
+      "loss": 0.5532,
+      "step": 3795
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.927552925674176e-06,
+      "loss": 0.5088,
+      "step": 3796
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.922697002768899e-06,
+      "loss": 0.5665,
+      "step": 3797
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9178433510531964e-06,
+      "loss": 0.524,
+      "step": 3798
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9129919723409635e-06,
+      "loss": 0.5344,
+      "step": 3799
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9081428684452335e-06,
+      "loss": 0.5406,
+      "step": 3800
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.903296041178203e-06,
+      "loss": 0.5256,
+      "step": 3801
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.898451492351209e-06,
+      "loss": 0.5365,
+      "step": 3802
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.893609223774743e-06,
+      "loss": 0.5496,
+      "step": 3803
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.888769237258441e-06,
+      "loss": 0.5115,
+      "step": 3804
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.883931534611085e-06,
+      "loss": 0.5562,
+      "step": 3805
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.879096117640612e-06,
+      "loss": 0.5382,
+      "step": 3806
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8742629881540875e-06,
+      "loss": 0.564,
+      "step": 3807
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.869432147957738e-06,
+      "loss": 0.5468,
+      "step": 3808
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.864603598856933e-06,
+      "loss": 0.5454,
+      "step": 3809
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8597773426561745e-06,
+      "loss": 0.5199,
+      "step": 3810
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.854953381159117e-06,
+      "loss": 0.5126,
+      "step": 3811
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.850131716168562e-06,
+      "loss": 0.4924,
+      "step": 3812
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8453123494864375e-06,
+      "loss": 0.5314,
+      "step": 3813
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.840495282913824e-06,
+      "loss": 0.5138,
+      "step": 3814
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8356805182509405e-06,
+      "loss": 0.5295,
+      "step": 3815
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.830868057297145e-06,
+      "loss": 0.516,
+      "step": 3816
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8260579018509345e-06,
+      "loss": 0.4799,
+      "step": 3817
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.821250053709944e-06,
+      "loss": 0.5353,
+      "step": 3818
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.816444514670951e-06,
+      "loss": 0.5522,
+      "step": 3819
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.811641286529857e-06,
+      "loss": 0.5371,
+      "step": 3820
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8068403710817126e-06,
+      "loss": 0.5043,
+      "step": 3821
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.802041770120706e-06,
+      "loss": 0.4872,
+      "step": 3822
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.797245485440143e-06,
+      "loss": 0.4698,
+      "step": 3823
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.792451518832483e-06,
+      "loss": 0.5596,
+      "step": 3824
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.787659872089312e-06,
+      "loss": 0.5638,
+      "step": 3825
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7828705470013428e-06,
+      "loss": 0.4952,
+      "step": 3826
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.77808354535843e-06,
+      "loss": 0.5272,
+      "step": 3827
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7732988689495575e-06,
+      "loss": 0.5325,
+      "step": 3828
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.768516519562834e-06,
+      "loss": 0.5172,
+      "step": 3829
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.763736498985505e-06,
+      "loss": 0.5328,
+      "step": 3830
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7589588090039454e-06,
+      "loss": 0.5426,
+      "step": 3831
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7541834514036567e-06,
+      "loss": 0.5306,
+      "step": 3832
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.749410427969271e-06,
+      "loss": 0.5376,
+      "step": 3833
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7446397404845446e-06,
+      "loss": 0.583,
+      "step": 3834
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.739871390732369e-06,
+      "loss": 0.5018,
+      "step": 3835
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7351053804947478e-06,
+      "loss": 0.5416,
+      "step": 3836
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.730341711552822e-06,
+      "loss": 0.5408,
+      "step": 3837
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7255803856868576e-06,
+      "loss": 0.5658,
+      "step": 3838
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7208214046762357e-06,
+      "loss": 0.5391,
+      "step": 3839
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.71606477029947e-06,
+      "loss": 0.5581,
+      "step": 3840
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7113104843341963e-06,
+      "loss": 0.5337,
+      "step": 3841
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7065585485571653e-06,
+      "loss": 0.5337,
+      "step": 3842
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.701808964744258e-06,
+      "loss": 0.5476,
+      "step": 3843
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6970617346704753e-06,
+      "loss": 0.5366,
+      "step": 3844
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.69231686010993e-06,
+      "loss": 0.5386,
+      "step": 3845
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6875743428358656e-06,
+      "loss": 0.5455,
+      "step": 3846
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6828341846206393e-06,
+      "loss": 0.5151,
+      "step": 3847
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.678096387235727e-06,
+      "loss": 0.5529,
+      "step": 3848
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6733609524517223e-06,
+      "loss": 0.4991,
+      "step": 3849
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.668627882038337e-06,
+      "loss": 0.5308,
+      "step": 3850
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6638971777644015e-06,
+      "loss": 0.5598,
+      "step": 3851
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6591688413978522e-06,
+      "loss": 0.5648,
+      "step": 3852
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6544428747057512e-06,
+      "loss": 0.5547,
+      "step": 3853
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6497192794542736e-06,
+      "loss": 0.5692,
+      "step": 3854
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6449980574086996e-06,
+      "loss": 0.5121,
+      "step": 3855
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6402792103334327e-06,
+      "loss": 0.5223,
+      "step": 3856
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.635562739991987e-06,
+      "loss": 0.5693,
+      "step": 3857
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.630848648146981e-06,
+      "loss": 0.5044,
+      "step": 3858
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6261369365601506e-06,
+      "loss": 0.5579,
+      "step": 3859
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6214276069923424e-06,
+      "loss": 0.5157,
+      "step": 3860
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6167206612035123e-06,
+      "loss": 0.5312,
+      "step": 3861
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.612016100952722e-06,
+      "loss": 0.5389,
+      "step": 3862
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.607313927998151e-06,
+      "loss": 0.5515,
+      "step": 3863
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.602614144097071e-06,
+      "loss": 0.5162,
+      "step": 3864
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5979167510058723e-06,
+      "loss": 0.5305,
+      "step": 3865
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5932217504800505e-06,
+      "loss": 0.5087,
+      "step": 3866
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5885291442742086e-06,
+      "loss": 0.5435,
+      "step": 3867
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5838389341420456e-06,
+      "loss": 0.5785,
+      "step": 3868
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.579151121836374e-06,
+      "loss": 0.5311,
+      "step": 3869
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5744657091091117e-06,
+      "loss": 0.502,
+      "step": 3870
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5697826977112683e-06,
+      "loss": 0.5267,
+      "step": 3871
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5651020893929668e-06,
+      "loss": 0.5641,
+      "step": 3872
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5604238859034313e-06,
+      "loss": 0.5292,
+      "step": 3873
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.555748088990979e-06,
+      "loss": 0.5063,
+      "step": 3874
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5510747004030364e-06,
+      "loss": 0.5189,
+      "step": 3875
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.546403721886126e-06,
+      "loss": 0.4995,
+      "step": 3876
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5417351551858713e-06,
+      "loss": 0.542,
+      "step": 3877
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5370690020469934e-06,
+      "loss": 0.5472,
+      "step": 3878
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5324052642133145e-06,
+      "loss": 0.5319,
+      "step": 3879
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.527743943427745e-06,
+      "loss": 0.5281,
+      "step": 3880
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5230850414323015e-06,
+      "loss": 0.5338,
+      "step": 3881
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5184285599680947e-06,
+      "loss": 0.5209,
+      "step": 3882
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.51377450077533e-06,
+      "loss": 0.5266,
+      "step": 3883
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5091228655933028e-06,
+      "loss": 0.5497,
+      "step": 3884
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5044736561604096e-06,
+      "loss": 0.5287,
+      "step": 3885
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4998268742141407e-06,
+      "loss": 0.5359,
+      "step": 3886
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4951825214910682e-06,
+      "loss": 0.5244,
+      "step": 3887
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4905405997268693e-06,
+      "loss": 0.5056,
+      "step": 3888
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4859011106563103e-06,
+      "loss": 0.5779,
+      "step": 3889
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.481264056013239e-06,
+      "loss": 0.561,
+      "step": 3890
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.476629437530604e-06,
+      "loss": 0.5706,
+      "step": 3891
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4719972569404393e-06,
+      "loss": 0.507,
+      "step": 3892
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4673675159738673e-06,
+      "loss": 0.5298,
+      "step": 3893
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4627402163611012e-06,
+      "loss": 0.5154,
+      "step": 3894
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.458115359831441e-06,
+      "loss": 0.5382,
+      "step": 3895
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4534929481132685e-06,
+      "loss": 0.5229,
+      "step": 3896
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4488729829340586e-06,
+      "loss": 0.5181,
+      "step": 3897
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.444255466020369e-06,
+      "loss": 0.5446,
+      "step": 3898
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4396403990978456e-06,
+      "loss": 0.5421,
+      "step": 3899
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4350277838912115e-06,
+      "loss": 0.5026,
+      "step": 3900
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.430417622124279e-06,
+      "loss": 0.5175,
+      "step": 3901
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4258099155199477e-06,
+      "loss": 0.4996,
+      "step": 3902
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4212046658001864e-06,
+      "loss": 0.5569,
+      "step": 3903
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4166018746860576e-06,
+      "loss": 0.5292,
+      "step": 3904
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4120015438977007e-06,
+      "loss": 0.5238,
+      "step": 3905
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.407403675154337e-06,
+      "loss": 0.4958,
+      "step": 3906
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4028082701742673e-06,
+      "loss": 0.5198,
+      "step": 3907
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.398215330674872e-06,
+      "loss": 0.5168,
+      "step": 3908
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.393624858372606e-06,
+      "loss": 0.5089,
+      "step": 3909
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3890368549830056e-06,
+      "loss": 0.4844,
+      "step": 3910
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.38445132222069e-06,
+      "loss": 0.5536,
+      "step": 3911
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3798682617993407e-06,
+      "loss": 0.5457,
+      "step": 3912
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3752876754317276e-06,
+      "loss": 0.531,
+      "step": 3913
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.370709564829694e-06,
+      "loss": 0.6035,
+      "step": 3914
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3661339317041573e-06,
+      "loss": 0.5323,
+      "step": 3915
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3615607777651037e-06,
+      "loss": 0.5233,
+      "step": 3916
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.356990104721599e-06,
+      "loss": 0.5345,
+      "step": 3917
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.352421914281783e-06,
+      "loss": 0.5314,
+      "step": 3918
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3478562081528587e-06,
+      "loss": 0.5112,
+      "step": 3919
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.34329298804111e-06,
+      "loss": 0.5352,
+      "step": 3920
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3387322556518877e-06,
+      "loss": 0.5239,
+      "step": 3921
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3341740126896148e-06,
+      "loss": 0.5509,
+      "step": 3922
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.329618260857783e-06,
+      "loss": 0.5521,
+      "step": 3923
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3250650018589547e-06,
+      "loss": 0.5262,
+      "step": 3924
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.320514237394753e-06,
+      "loss": 0.5488,
+      "step": 3925
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3159659691658776e-06,
+      "loss": 0.5624,
+      "step": 3926
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.311420198872095e-06,
+      "loss": 0.5531,
+      "step": 3927
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3068769282122302e-06,
+      "loss": 0.4723,
+      "step": 3928
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.302336158884182e-06,
+      "loss": 0.5599,
+      "step": 3929
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.297797892584912e-06,
+      "loss": 0.5015,
+      "step": 3930
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2932621310104484e-06,
+      "loss": 0.5726,
+      "step": 3931
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2887288758558745e-06,
+      "loss": 0.5278,
+      "step": 3932
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.284198128815348e-06,
+      "loss": 0.5056,
+      "step": 3933
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.279669891582087e-06,
+      "loss": 0.5304,
+      "step": 3934
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2751441658483617e-06,
+      "loss": 0.5544,
+      "step": 3935
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2706209533055178e-06,
+      "loss": 0.495,
+      "step": 3936
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2661002556439527e-06,
+      "loss": 0.5635,
+      "step": 3937
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2615820745531267e-06,
+      "loss": 0.5303,
+      "step": 3938
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.25706641172156e-06,
+      "loss": 0.5045,
+      "step": 3939
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2525532688368354e-06,
+      "loss": 0.5715,
+      "step": 3940
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.248042647585582e-06,
+      "loss": 0.4873,
+      "step": 3941
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2435345496534976e-06,
+      "loss": 0.5682,
+      "step": 3942
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.239028976725337e-06,
+      "loss": 0.5082,
+      "step": 3943
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2345259304849017e-06,
+      "loss": 0.5259,
+      "step": 3944
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2300254126150597e-06,
+      "loss": 0.5192,
+      "step": 3945
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2255274247977274e-06,
+      "loss": 0.573,
+      "step": 3946
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2210319687138824e-06,
+      "loss": 0.5438,
+      "step": 3947
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2165390460435443e-06,
+      "loss": 0.4705,
+      "step": 3948
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2120486584657974e-06,
+      "loss": 0.5125,
+      "step": 3949
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.207560807658775e-06,
+      "loss": 0.5211,
+      "step": 3950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2030754952996612e-06,
+      "loss": 0.5492,
+      "step": 3951
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1985927230646906e-06,
+      "loss": 0.5354,
+      "step": 3952
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1941124926291566e-06,
+      "loss": 0.5492,
+      "step": 3953
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1896348056673864e-06,
+      "loss": 0.522,
+      "step": 3954
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1851596638527715e-06,
+      "loss": 0.5488,
+      "step": 3955
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1806870688577496e-06,
+      "loss": 0.4789,
+      "step": 3956
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.176217022353798e-06,
+      "loss": 0.5341,
+      "step": 3957
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1717495260114505e-06,
+      "loss": 0.5222,
+      "step": 3958
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1672845815002886e-06,
+      "loss": 0.4996,
+      "step": 3959
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.16282219048893e-06,
+      "loss": 0.4988,
+      "step": 3960
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1583623546450483e-06,
+      "loss": 0.5225,
+      "step": 3961
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1539050756353584e-06,
+      "loss": 0.5076,
+      "step": 3962
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1494503551256227e-06,
+      "loss": 0.5112,
+      "step": 3963
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1449981947806387e-06,
+      "loss": 0.5164,
+      "step": 3964
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1405485962642556e-06,
+      "loss": 0.5085,
+      "step": 3965
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.136101561239363e-06,
+      "loss": 0.5294,
+      "step": 3966
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.131657091367892e-06,
+      "loss": 0.6051,
+      "step": 3967
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.127215188310817e-06,
+      "loss": 0.5297,
+      "step": 3968
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.122775853728152e-06,
+      "loss": 0.501,
+      "step": 3969
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1183390892789433e-06,
+      "loss": 0.5591,
+      "step": 3970
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1139048966212905e-06,
+      "loss": 0.5271,
+      "step": 3971
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.109473277412326e-06,
+      "loss": 0.4935,
+      "step": 3972
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1050442333082143e-06,
+      "loss": 0.5568,
+      "step": 3973
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1006177659641666e-06,
+      "loss": 0.5066,
+      "step": 3974
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.09619387703443e-06,
+      "loss": 0.5315,
+      "step": 3975
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0917725681722808e-06,
+      "loss": 0.5082,
+      "step": 3976
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0873538410300375e-06,
+      "loss": 0.514,
+      "step": 3977
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.082937697259052e-06,
+      "loss": 0.5577,
+      "step": 3978
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.078524138509715e-06,
+      "loss": 0.5282,
+      "step": 3979
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0741131664314418e-06,
+      "loss": 0.5154,
+      "step": 3980
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0697047826726867e-06,
+      "loss": 0.517,
+      "step": 3981
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.065298988880939e-06,
+      "loss": 0.5724,
+      "step": 3982
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0608957867027154e-06,
+      "loss": 0.5476,
+      "step": 3983
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0564951777835674e-06,
+      "loss": 0.5267,
+      "step": 3984
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0520971637680784e-06,
+      "loss": 0.521,
+      "step": 3985
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0477017462998537e-06,
+      "loss": 0.5155,
+      "step": 3986
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0433089270215375e-06,
+      "loss": 0.5471,
+      "step": 3987
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.038918707574803e-06,
+      "loss": 0.5244,
+      "step": 3988
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.034531089600341e-06,
+      "loss": 0.5526,
+      "step": 3989
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.030146074737882e-06,
+      "loss": 0.4886,
+      "step": 3990
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.025763664626181e-06,
+      "loss": 0.5356,
+      "step": 3991
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0213838609030132e-06,
+      "loss": 0.5258,
+      "step": 3992
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0170066652051865e-06,
+      "loss": 0.5389,
+      "step": 3993
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0126320791685314e-06,
+      "loss": 0.5361,
+      "step": 3994
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.008260104427906e-06,
+      "loss": 0.5157,
+      "step": 3995
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0038907426171883e-06,
+      "loss": 0.532,
+      "step": 3996
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.999523995369281e-06,
+      "loss": 0.5605,
+      "step": 3997
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9951598643161163e-06,
+      "loss": 0.5543,
+      "step": 3998
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.990798351088635e-06,
+      "loss": 0.5154,
+      "step": 3999
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.98643945731681e-06,
+      "loss": 0.5731,
+      "step": 4000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9820831846296362e-06,
+      "loss": 0.524,
+      "step": 4001
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9777295346551195e-06,
+      "loss": 0.5217,
+      "step": 4002
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.973378509020295e-06,
+      "loss": 0.5353,
+      "step": 4003
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.969030109351215e-06,
+      "loss": 0.5411,
+      "step": 4004
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.964684337272946e-06,
+      "loss": 0.5088,
+      "step": 4005
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.960341194409575e-06,
+      "loss": 0.5186,
+      "step": 4006
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9560006823842136e-06,
+      "loss": 0.5482,
+      "step": 4007
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.951662802818974e-06,
+      "loss": 0.5234,
+      "step": 4008
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9473275573350003e-06,
+      "loss": 0.5231,
+      "step": 4009
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9429949475524434e-06,
+      "loss": 0.5229,
+      "step": 4010
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9386649750904737e-06,
+      "loss": 0.5018,
+      "step": 4011
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.934337641567274e-06,
+      "loss": 0.5013,
+      "step": 4012
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9300129486000405e-06,
+      "loss": 0.5034,
+      "step": 4013
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9256908978049868e-06,
+      "loss": 0.4995,
+      "step": 4014
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.921371490797329e-06,
+      "loss": 0.5267,
+      "step": 4015
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.917054729191304e-06,
+      "loss": 0.5679,
+      "step": 4016
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9127406146001625e-06,
+      "loss": 0.5089,
+      "step": 4017
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9084291486361538e-06,
+      "loss": 0.4857,
+      "step": 4018
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.904120332910547e-06,
+      "loss": 0.5039,
+      "step": 4019
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8998141690336223e-06,
+      "loss": 0.4993,
+      "step": 4020
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8955106586146596e-06,
+      "loss": 0.5256,
+      "step": 4021
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.891209803261955e-06,
+      "loss": 0.5138,
+      "step": 4022
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8869116045828103e-06,
+      "loss": 0.5521,
+      "step": 4023
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8826160641835334e-06,
+      "loss": 0.5195,
+      "step": 4024
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.87832318366944e-06,
+      "loss": 0.549,
+      "step": 4025
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8740329646448505e-06,
+      "loss": 0.5322,
+      "step": 4026
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.869745408713096e-06,
+      "loss": 0.5531,
+      "step": 4027
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8654605174764994e-06,
+      "loss": 0.5215,
+      "step": 4028
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.861178292536401e-06,
+      "loss": 0.5145,
+      "step": 4029
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.856898735493142e-06,
+      "loss": 0.5482,
+      "step": 4030
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8526218479460576e-06,
+      "loss": 0.5355,
+      "step": 4031
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8483476314934966e-06,
+      "loss": 0.5632,
+      "step": 4032
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8440760877328078e-06,
+      "loss": 0.5411,
+      "step": 4033
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.839807218260331e-06,
+      "loss": 0.543,
+      "step": 4034
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8355410246714176e-06,
+      "loss": 0.5194,
+      "step": 4035
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8312775085604195e-06,
+      "loss": 0.5325,
+      "step": 4036
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.827016671520676e-06,
+      "loss": 0.5358,
+      "step": 4037
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8227585151445367e-06,
+      "loss": 0.5118,
+      "step": 4038
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8185030410233473e-06,
+      "loss": 0.4868,
+      "step": 4039
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.814250250747447e-06,
+      "loss": 0.5079,
+      "step": 4040
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8100001459061764e-06,
+      "loss": 0.4937,
+      "step": 4041
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8057527280878694e-06,
+      "loss": 0.5485,
+      "step": 4042
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.801507998879861e-06,
+      "loss": 0.5232,
+      "step": 4043
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7972659598684703e-06,
+      "loss": 0.4943,
+      "step": 4044
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7930266126390217e-06,
+      "loss": 0.5151,
+      "step": 4045
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7887899587758327e-06,
+      "loss": 0.5065,
+      "step": 4046
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7845559998622064e-06,
+      "loss": 0.5244,
+      "step": 4047
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7803247374804465e-06,
+      "loss": 0.5397,
+      "step": 4048
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7760961732118484e-06,
+      "loss": 0.5233,
+      "step": 4049
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7718703086366926e-06,
+      "loss": 0.5551,
+      "step": 4050
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7676471453342568e-06,
+      "loss": 0.5428,
+      "step": 4051
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.763426684882812e-06,
+      "loss": 0.541,
+      "step": 4052
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.759208928859608e-06,
+      "loss": 0.5701,
+      "step": 4053
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7549938788408936e-06,
+      "loss": 0.4983,
+      "step": 4054
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7507815364019044e-06,
+      "loss": 0.5436,
+      "step": 4055
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7465719031168612e-06,
+      "loss": 0.4917,
+      "step": 4056
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.742364980558976e-06,
+      "loss": 0.5532,
+      "step": 4057
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7381607703004444e-06,
+      "loss": 0.5743,
+      "step": 4058
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7339592739124543e-06,
+      "loss": 0.4996,
+      "step": 4059
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7297604929651677e-06,
+      "loss": 0.5383,
+      "step": 4060
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.725564429027744e-06,
+      "loss": 0.5093,
+      "step": 4061
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.721371083668324e-06,
+      "loss": 0.4963,
+      "step": 4062
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.717180458454025e-06,
+      "loss": 0.5393,
+      "step": 4063
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7129925549509563e-06,
+      "loss": 0.535,
+      "step": 4064
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7088073747242107e-06,
+      "loss": 0.5117,
+      "step": 4065
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.704624919337855e-06,
+      "loss": 0.5476,
+      "step": 4066
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7004451903549435e-06,
+      "loss": 0.5105,
+      "step": 4067
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.696268189337512e-06,
+      "loss": 0.5082,
+      "step": 4068
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6920939178465754e-06,
+      "loss": 0.5084,
+      "step": 4069
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6879223774421282e-06,
+      "loss": 0.5304,
+      "step": 4070
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6837535696831487e-06,
+      "loss": 0.545,
+      "step": 4071
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6795874961275823e-06,
+      "loss": 0.5054,
+      "step": 4072
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6754241583323647e-06,
+      "loss": 0.4967,
+      "step": 4073
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.671263557853404e-06,
+      "loss": 0.5078,
+      "step": 4074
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6671056962455898e-06,
+      "loss": 0.5303,
+      "step": 4075
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.662950575062777e-06,
+      "loss": 0.5498,
+      "step": 4076
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.658798195857808e-06,
+      "loss": 0.5178,
+      "step": 4077
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.654648560182499e-06,
+      "loss": 0.5157,
+      "step": 4078
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.650501669587632e-06,
+      "loss": 0.5873,
+      "step": 4079
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6463575256229713e-06,
+      "loss": 0.5431,
+      "step": 4080
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6422161298372563e-06,
+      "loss": 0.5763,
+      "step": 4081
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6380774837781907e-06,
+      "loss": 0.5206,
+      "step": 4082
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.633941588992458e-06,
+      "loss": 0.4733,
+      "step": 4083
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.62980844702571e-06,
+      "loss": 0.4971,
+      "step": 4084
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6256780594225727e-06,
+      "loss": 0.5353,
+      "step": 4085
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6215504277266402e-06,
+      "loss": 0.5284,
+      "step": 4086
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6174255534804803e-06,
+      "loss": 0.5256,
+      "step": 4087
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.613303438225621e-06,
+      "loss": 0.5057,
+      "step": 4088
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6091840835025683e-06,
+      "loss": 0.554,
+      "step": 4089
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6050674908507945e-06,
+      "loss": 0.5311,
+      "step": 4090
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6009536618087417e-06,
+      "loss": 0.5188,
+      "step": 4091
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.596842597913811e-06,
+      "loss": 0.5236,
+      "step": 4092
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5927343007023784e-06,
+      "loss": 0.5483,
+      "step": 4093
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5886287717097856e-06,
+      "loss": 0.5228,
+      "step": 4094
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.584526012470332e-06,
+      "loss": 0.532,
+      "step": 4095
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5804260245172896e-06,
+      "loss": 0.5432,
+      "step": 4096
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5763288093828966e-06,
+      "loss": 0.5094,
+      "step": 4097
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.572234368598342e-06,
+      "loss": 0.5551,
+      "step": 4098
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5681427036937924e-06,
+      "loss": 0.5675,
+      "step": 4099
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5640538161983708e-06,
+      "loss": 0.52,
+      "step": 4100
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5599677076401607e-06,
+      "loss": 0.5168,
+      "step": 4101
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.555884379546212e-06,
+      "loss": 0.5249,
+      "step": 4102
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5518038334425333e-06,
+      "loss": 0.5065,
+      "step": 4103
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5477260708540876e-06,
+      "loss": 0.5189,
+      "step": 4104
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5436510933048063e-06,
+      "loss": 0.5543,
+      "step": 4105
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5395789023175752e-06,
+      "loss": 0.5505,
+      "step": 4106
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.535509499414245e-06,
+      "loss": 0.5256,
+      "step": 4107
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5314428861156127e-06,
+      "loss": 0.5642,
+      "step": 4108
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.527379063941441e-06,
+      "loss": 0.5114,
+      "step": 4109
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5233180344104536e-06,
+      "loss": 0.5581,
+      "step": 4110
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.519259799040317e-06,
+      "loss": 0.5298,
+      "step": 4111
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.515204359347665e-06,
+      "loss": 0.5013,
+      "step": 4112
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5111517168480837e-06,
+      "loss": 0.5397,
+      "step": 4113
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.507101873056114e-06,
+      "loss": 0.5048,
+      "step": 4114
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5030548294852484e-06,
+      "loss": 0.5188,
+      "step": 4115
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4990105876479375e-06,
+      "loss": 0.5103,
+      "step": 4116
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.494969149055578e-06,
+      "loss": 0.5369,
+      "step": 4117
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4909305152185248e-06,
+      "loss": 0.5236,
+      "step": 4118
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.486894687646085e-06,
+      "loss": 0.5153,
+      "step": 4119
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.48286166784651e-06,
+      "loss": 0.4963,
+      "step": 4120
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.478831457327009e-06,
+      "loss": 0.5034,
+      "step": 4121
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.474804057593738e-06,
+      "loss": 0.5438,
+      "step": 4122
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4707794701518095e-06,
+      "loss": 0.5212,
+      "step": 4123
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4667576965052697e-06,
+      "loss": 0.5221,
+      "step": 4124
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.462738738157128e-06,
+      "loss": 0.5229,
+      "step": 4125
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4587225966093365e-06,
+      "loss": 0.5054,
+      "step": 4126
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.454709273362791e-06,
+      "loss": 0.532,
+      "step": 4127
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.450698769917339e-06,
+      "loss": 0.5231,
+      "step": 4128
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.446691087771773e-06,
+      "loss": 0.526,
+      "step": 4129
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.442686228423832e-06,
+      "loss": 0.5431,
+      "step": 4130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4386841933701955e-06,
+      "loss": 0.5374,
+      "step": 4131
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.434684984106497e-06,
+      "loss": 0.5104,
+      "step": 4132
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4306886021273e-06,
+      "loss": 0.5332,
+      "step": 4133
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.426695048926123e-06,
+      "loss": 0.5225,
+      "step": 4134
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.422704325995425e-06,
+      "loss": 0.5161,
+      "step": 4135
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4187164348266024e-06,
+      "loss": 0.5137,
+      "step": 4136
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.414731376909998e-06,
+      "loss": 0.536,
+      "step": 4137
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4107491537348938e-06,
+      "loss": 0.5132,
+      "step": 4138
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4067697667895175e-06,
+      "loss": 0.5424,
+      "step": 4139
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.402793217561026e-06,
+      "loss": 0.5564,
+      "step": 4140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.398819507535526e-06,
+      "loss": 0.5344,
+      "step": 4141
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3948486381980596e-06,
+      "loss": 0.514,
+      "step": 4142
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.390880611032603e-06,
+      "loss": 0.5164,
+      "step": 4143
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3869154275220774e-06,
+      "loss": 0.4965,
+      "step": 4144
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3829530891483378e-06,
+      "loss": 0.5305,
+      "step": 4145
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3789935973921754e-06,
+      "loss": 0.5314,
+      "step": 4146
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3750369537333186e-06,
+      "loss": 0.5258,
+      "step": 4147
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3710831596504337e-06,
+      "loss": 0.5735,
+      "step": 4148
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3671322166211154e-06,
+      "loss": 0.542,
+      "step": 4149
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.363184126121899e-06,
+      "loss": 0.5387,
+      "step": 4150
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.359238889628254e-06,
+      "loss": 0.5305,
+      "step": 4151
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3552965086145764e-06,
+      "loss": 0.5237,
+      "step": 4152
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.351356984554203e-06,
+      "loss": 0.5056,
+      "step": 4153
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.347420318919398e-06,
+      "loss": 0.5143,
+      "step": 4154
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3434865131813635e-06,
+      "loss": 0.5341,
+      "step": 4155
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.4955,
+      "step": 4156
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.335627487275035e-06,
+      "loss": 0.5272,
+      "step": 4157
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.331702270043793e-06,
+      "loss": 0.5146,
+      "step": 4158
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.327779918583417e-06,
+      "loss": 0.5342,
+      "step": 4159
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3238604343597524e-06,
+      "loss": 0.522,
+      "step": 4160
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.31994381883758e-06,
+      "loss": 0.5083,
+      "step": 4161
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3160300734805985e-06,
+      "loss": 0.5618,
+      "step": 4162
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3121191997514425e-06,
+      "loss": 0.4863,
+      "step": 4163
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.308211199111674e-06,
+      "loss": 0.5135,
+      "step": 4164
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3043060730217726e-06,
+      "loss": 0.5562,
+      "step": 4165
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.300403822941152e-06,
+      "loss": 0.553,
+      "step": 4166
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2965044503281497e-06,
+      "loss": 0.5149,
+      "step": 4167
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.292607956640022e-06,
+      "loss": 0.527,
+      "step": 4168
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.288714343332956e-06,
+      "loss": 0.5347,
+      "step": 4169
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2848236118620605e-06,
+      "loss": 0.5073,
+      "step": 4170
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.280935763681369e-06,
+      "loss": 0.535,
+      "step": 4171
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2770508002438295e-06,
+      "loss": 0.5525,
+      "step": 4172
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.273168723001321e-06,
+      "loss": 0.5159,
+      "step": 4173
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2692895334046404e-06,
+      "loss": 0.5171,
+      "step": 4174
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.265413232903506e-06,
+      "loss": 0.4888,
+      "step": 4175
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2615398229465548e-06,
+      "loss": 0.5061,
+      "step": 4176
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.257669304981349e-06,
+      "loss": 0.5233,
+      "step": 4177
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2538016804543596e-06,
+      "loss": 0.5432,
+      "step": 4178
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2499369508109847e-06,
+      "loss": 0.5176,
+      "step": 4179
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.246075117495542e-06,
+      "loss": 0.5526,
+      "step": 4180
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2422161819512566e-06,
+      "loss": 0.5189,
+      "step": 4181
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.23836014562028e-06,
+      "loss": 0.5123,
+      "step": 4182
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.234507009943682e-06,
+      "loss": 0.5312,
+      "step": 4183
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.230656776361435e-06,
+      "loss": 0.4968,
+      "step": 4184
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.22680944631244e-06,
+      "loss": 0.5148,
+      "step": 4185
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.222965021234509e-06,
+      "loss": 0.5179,
+      "step": 4186
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.21912350256437e-06,
+      "loss": 0.579,
+      "step": 4187
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.215284891737658e-06,
+      "loss": 0.5339,
+      "step": 4188
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.211449190188927e-06,
+      "loss": 0.5166,
+      "step": 4189
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.207616399351643e-06,
+      "loss": 0.5465,
+      "step": 4190
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2037865206581855e-06,
+      "loss": 0.5469,
+      "step": 4191
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1999595555398425e-06,
+      "loss": 0.4608,
+      "step": 4192
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1961355054268173e-06,
+      "loss": 0.5115,
+      "step": 4193
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1923143717482175e-06,
+      "loss": 0.5153,
+      "step": 4194
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.188496155932066e-06,
+      "loss": 0.519,
+      "step": 4195
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.184680859405297e-06,
+      "loss": 0.5302,
+      "step": 4196
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1808684835937454e-06,
+      "loss": 0.5802,
+      "step": 4197
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1770590299221616e-06,
+      "loss": 0.5608,
+      "step": 4198
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1732524998142047e-06,
+      "loss": 0.5232,
+      "step": 4199
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.169448894692433e-06,
+      "loss": 0.4973,
+      "step": 4200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.165648215978321e-06,
+      "loss": 0.5125,
+      "step": 4201
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1618504650922446e-06,
+      "loss": 0.5212,
+      "step": 4202
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.158055643453488e-06,
+      "loss": 0.4829,
+      "step": 4203
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1542637524802377e-06,
+      "loss": 0.5287,
+      "step": 4204
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1504747935895886e-06,
+      "loss": 0.493,
+      "step": 4205
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.146688768197539e-06,
+      "loss": 0.5494,
+      "step": 4206
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.142905677718985e-06,
+      "loss": 0.5331,
+      "step": 4207
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1391255235677335e-06,
+      "loss": 0.5198,
+      "step": 4208
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.135348307156494e-06,
+      "loss": 0.5614,
+      "step": 4209
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.131574029896869e-06,
+      "loss": 0.505,
+      "step": 4210
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.127802693199372e-06,
+      "loss": 0.5509,
+      "step": 4211
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1240342984734186e-06,
+      "loss": 0.5459,
+      "step": 4212
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1202688471273137e-06,
+      "loss": 0.5177,
+      "step": 4213
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1165063405682715e-06,
+      "loss": 0.5143,
+      "step": 4214
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1127467802024082e-06,
+      "loss": 0.526,
+      "step": 4215
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.108990167434728e-06,
+      "loss": 0.5597,
+      "step": 4216
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1052365036691426e-06,
+      "loss": 0.5492,
+      "step": 4217
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1014857903084585e-06,
+      "loss": 0.5415,
+      "step": 4218
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0977380287543804e-06,
+      "loss": 0.5536,
+      "step": 4219
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0939932204075086e-06,
+      "loss": 0.5226,
+      "step": 4220
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.090251366667341e-06,
+      "loss": 0.5003,
+      "step": 4221
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0865124689322726e-06,
+      "loss": 0.5195,
+      "step": 4222
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.082776528599588e-06,
+      "loss": 0.5287,
+      "step": 4223
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0790435470654723e-06,
+      "loss": 0.4918,
+      "step": 4224
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.075313525725007e-06,
+      "loss": 0.559,
+      "step": 4225
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.071586465972155e-06,
+      "loss": 0.5156,
+      "step": 4226
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0678623691997858e-06,
+      "loss": 0.4974,
+      "step": 4227
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.064141236799659e-06,
+      "loss": 0.4855,
+      "step": 4228
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0604230701624184e-06,
+      "loss": 0.5431,
+      "step": 4229
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0567078706776066e-06,
+      "loss": 0.481,
+      "step": 4230
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0529956397336593e-06,
+      "loss": 0.4996,
+      "step": 4231
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0492863787178942e-06,
+      "loss": 0.5107,
+      "step": 4232
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0455800890165256e-06,
+      "loss": 0.495,
+      "step": 4233
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.041876772014657e-06,
+      "loss": 0.5488,
+      "step": 4234
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.038176429096279e-06,
+      "loss": 0.5179,
+      "step": 4235
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.034479061644271e-06,
+      "loss": 0.4772,
+      "step": 4236
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0307846710404023e-06,
+      "loss": 0.4993,
+      "step": 4237
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0270932586653303e-06,
+      "loss": 0.5293,
+      "step": 4238
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0234048258985927e-06,
+      "loss": 0.5565,
+      "step": 4239
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0197193741186194e-06,
+      "loss": 0.4918,
+      "step": 4240
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0160369047027293e-06,
+      "loss": 0.5336,
+      "step": 4241
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.012357419027118e-06,
+      "loss": 0.5247,
+      "step": 4242
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0086809184668722e-06,
+      "loss": 0.5154,
+      "step": 4243
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0050074043959634e-06,
+      "loss": 0.5299,
+      "step": 4244
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0013368781872412e-06,
+      "loss": 0.5132,
+      "step": 4245
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.997669341212446e-06,
+      "loss": 0.5092,
+      "step": 4246
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.994004794842195e-06,
+      "loss": 0.5474,
+      "step": 4247
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.990343240445992e-06,
+      "loss": 0.549,
+      "step": 4248
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.986684679392221e-06,
+      "loss": 0.5444,
+      "step": 4249
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.983029113048147e-06,
+      "loss": 0.5181,
+      "step": 4250
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9793765427799194e-06,
+      "loss": 0.5314,
+      "step": 4251
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9757269699525596e-06,
+      "loss": 0.5708,
+      "step": 4252
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.972080395929974e-06,
+      "loss": 0.5348,
+      "step": 4253
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.968436822074954e-06,
+      "loss": 0.5135,
+      "step": 4254
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9647962497491567e-06,
+      "loss": 0.5468,
+      "step": 4255
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9611586803131277e-06,
+      "loss": 0.5314,
+      "step": 4256
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9575241151262916e-06,
+      "loss": 0.5256,
+      "step": 4257
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.953892555546938e-06,
+      "loss": 0.5315,
+      "step": 4258
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.950264002932246e-06,
+      "loss": 0.524,
+      "step": 4259
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.946638458638269e-06,
+      "loss": 0.561,
+      "step": 4260
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9430159240199277e-06,
+      "loss": 0.5046,
+      "step": 4261
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9393964004310264e-06,
+      "loss": 0.5303,
+      "step": 4262
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.935779889224241e-06,
+      "loss": 0.5215,
+      "step": 4263
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9321663917511226e-06,
+      "loss": 0.5206,
+      "step": 4264
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9285559093620952e-06,
+      "loss": 0.5337,
+      "step": 4265
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.924948443406457e-06,
+      "loss": 0.5245,
+      "step": 4266
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.921343995232381e-06,
+      "loss": 0.5002,
+      "step": 4267
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.917742566186902e-06,
+      "loss": 0.5552,
+      "step": 4268
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9141441576159403e-06,
+      "loss": 0.5298,
+      "step": 4269
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9105487708642813e-06,
+      "loss": 0.5118,
+      "step": 4270
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9069564072755776e-06,
+      "loss": 0.5171,
+      "step": 4271
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9033670681923566e-06,
+      "loss": 0.5479,
+      "step": 4272
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8997807549560165e-06,
+      "loss": 0.5249,
+      "step": 4273
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8961974689068196e-06,
+      "loss": 0.5427,
+      "step": 4274
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8926172113839003e-06,
+      "loss": 0.5266,
+      "step": 4275
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8890399837252638e-06,
+      "loss": 0.5227,
+      "step": 4276
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8854657872677739e-06,
+      "loss": 0.5363,
+      "step": 4277
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8818946233471713e-06,
+      "loss": 0.5366,
+      "step": 4278
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8783264932980583e-06,
+      "loss": 0.4939,
+      "step": 4279
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8747613984539059e-06,
+      "loss": 0.5011,
+      "step": 4280
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8711993401470484e-06,
+      "loss": 0.5323,
+      "step": 4281
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8676403197086879e-06,
+      "loss": 0.5767,
+      "step": 4282
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8640843384688901e-06,
+      "loss": 0.5338,
+      "step": 4283
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.860531397756582e-06,
+      "loss": 0.5071,
+      "step": 4284
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8569814988995572e-06,
+      "loss": 0.5317,
+      "step": 4285
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8534346432244765e-06,
+      "loss": 0.5266,
+      "step": 4286
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8498908320568531e-06,
+      "loss": 0.5503,
+      "step": 4287
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8463500667210698e-06,
+      "loss": 0.5425,
+      "step": 4288
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.842812348540375e-06,
+      "loss": 0.4854,
+      "step": 4289
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.839277678836866e-06,
+      "loss": 0.537,
+      "step": 4290
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8357460589315101e-06,
+      "loss": 0.5468,
+      "step": 4291
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8322174901441336e-06,
+      "loss": 0.5249,
+      "step": 4292
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8286919737934216e-06,
+      "loss": 0.5348,
+      "step": 4293
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.825169511196917e-06,
+      "loss": 0.579,
+      "step": 4294
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8216501036710278e-06,
+      "loss": 0.5288,
+      "step": 4295
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8181337525310084e-06,
+      "loss": 0.5404,
+      "step": 4296
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.814620459090981e-06,
+      "loss": 0.5176,
+      "step": 4297
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8111102246639224e-06,
+      "loss": 0.4949,
+      "step": 4298
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.807603050561667e-06,
+      "loss": 0.5153,
+      "step": 4299
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8040989380949003e-06,
+      "loss": 0.491,
+      "step": 4300
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8005978885731712e-06,
+      "loss": 0.4871,
+      "step": 4301
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.797099903304881e-06,
+      "loss": 0.5327,
+      "step": 4302
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7936049835972802e-06,
+      "loss": 0.5101,
+      "step": 4303
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7901131307564824e-06,
+      "loss": 0.5243,
+      "step": 4304
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7866243460874534e-06,
+      "loss": 0.5184,
+      "step": 4305
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7831386308940046e-06,
+      "loss": 0.5244,
+      "step": 4306
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7796559864788088e-06,
+      "loss": 0.5264,
+      "step": 4307
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.776176414143389e-06,
+      "loss": 0.5348,
+      "step": 4308
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7726999151881186e-06,
+      "loss": 0.5021,
+      "step": 4309
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7692264909122237e-06,
+      "loss": 0.568,
+      "step": 4310
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7657561426137836e-06,
+      "loss": 0.5508,
+      "step": 4311
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.762288871589719e-06,
+      "loss": 0.5062,
+      "step": 4312
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.758824679135812e-06,
+      "loss": 0.5304,
+      "step": 4313
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7553635665466873e-06,
+      "loss": 0.5892,
+      "step": 4314
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7519055351158233e-06,
+      "loss": 0.5332,
+      "step": 4315
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7484505861355394e-06,
+      "loss": 0.5029,
+      "step": 4316
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7449987208970088e-06,
+      "loss": 0.5134,
+      "step": 4317
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.741549940690256e-06,
+      "loss": 0.5062,
+      "step": 4318
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.738104246804141e-06,
+      "loss": 0.4985,
+      "step": 4319
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7346616405263804e-06,
+      "loss": 0.493,
+      "step": 4320
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7312221231435367e-06,
+      "loss": 0.5426,
+      "step": 4321
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7277856959410077e-06,
+      "loss": 0.528,
+      "step": 4322
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7243523602030488e-06,
+      "loss": 0.4934,
+      "step": 4323
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7209221172127533e-06,
+      "loss": 0.5073,
+      "step": 4324
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7174949682520603e-06,
+      "loss": 0.5022,
+      "step": 4325
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.714070914601752e-06,
+      "loss": 0.5468,
+      "step": 4326
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7106499575414593e-06,
+      "loss": 0.5111,
+      "step": 4327
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7072320983496437e-06,
+      "loss": 0.5502,
+      "step": 4328
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7038173383036205e-06,
+      "loss": 0.5498,
+      "step": 4329
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7004056786795397e-06,
+      "loss": 0.5027,
+      "step": 4330
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6969971207524017e-06,
+      "loss": 0.5055,
+      "step": 4331
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6935916657960328e-06,
+      "loss": 0.5265,
+      "step": 4332
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6901893150831139e-06,
+      "loss": 0.4978,
+      "step": 4333
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6867900698851615e-06,
+      "loss": 0.5304,
+      "step": 4334
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6833939314725245e-06,
+      "loss": 0.5271,
+      "step": 4335
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6800009011144003e-06,
+      "loss": 0.5443,
+      "step": 4336
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.67661098007882e-06,
+      "loss": 0.5389,
+      "step": 4337
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6732241696326544e-06,
+      "loss": 0.4869,
+      "step": 4338
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6698404710416105e-06,
+      "loss": 0.5044,
+      "step": 4339
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6664598855702363e-06,
+      "loss": 0.5203,
+      "step": 4340
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6630824144819069e-06,
+      "loss": 0.5192,
+      "step": 4341
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6597080590388425e-06,
+      "loss": 0.5509,
+      "step": 4342
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6563368205021002e-06,
+      "loss": 0.4845,
+      "step": 4343
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6529687001315597e-06,
+      "loss": 0.5076,
+      "step": 4344
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6496036991859498e-06,
+      "loss": 0.5325,
+      "step": 4345
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6462418189228258e-06,
+      "loss": 0.5034,
+      "step": 4346
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6428830605985801e-06,
+      "loss": 0.5317,
+      "step": 4347
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6395274254684346e-06,
+      "loss": 0.5645,
+      "step": 4348
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.636174914786446e-06,
+      "loss": 0.5375,
+      "step": 4349
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6328255298055063e-06,
+      "loss": 0.5321,
+      "step": 4350
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6294792717773322e-06,
+      "loss": 0.5603,
+      "step": 4351
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6261361419524802e-06,
+      "loss": 0.552,
+      "step": 4352
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6227961415803307e-06,
+      "loss": 0.5214,
+      "step": 4353
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6194592719090996e-06,
+      "loss": 0.5209,
+      "step": 4354
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6161255341858307e-06,
+      "loss": 0.5142,
+      "step": 4355
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6127949296563982e-06,
+      "loss": 0.5098,
+      "step": 4356
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6094674595655012e-06,
+      "loss": 0.5574,
+      "step": 4357
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.606143125156674e-06,
+      "loss": 0.5023,
+      "step": 4358
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6028219276722767e-06,
+      "loss": 0.5555,
+      "step": 4359
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5995038683534924e-06,
+      "loss": 0.5231,
+      "step": 4360
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.596188948440337e-06,
+      "loss": 0.5853,
+      "step": 4361
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5928771691716516e-06,
+      "loss": 0.5638,
+      "step": 4362
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5895685317851062e-06,
+      "loss": 0.4924,
+      "step": 4363
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5862630375171884e-06,
+      "loss": 0.5439,
+      "step": 4364
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5829606876032189e-06,
+      "loss": 0.4886,
+      "step": 4365
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5796614832773449e-06,
+      "loss": 0.5116,
+      "step": 4366
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.576365425772527e-06,
+      "loss": 0.5317,
+      "step": 4367
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5730725163205618e-06,
+      "loss": 0.5189,
+      "step": 4368
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5697827561520618e-06,
+      "loss": 0.4818,
+      "step": 4369
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5664961464964678e-06,
+      "loss": 0.5596,
+      "step": 4370
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5632126885820387e-06,
+      "loss": 0.5016,
+      "step": 4371
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5599323836358614e-06,
+      "loss": 0.5178,
+      "step": 4372
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5566552328838358e-06,
+      "loss": 0.5511,
+      "step": 4373
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5533812375506886e-06,
+      "loss": 0.5161,
+      "step": 4374
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5501103988599708e-06,
+      "loss": 0.5171,
+      "step": 4375
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5468427180340439e-06,
+      "loss": 0.4959,
+      "step": 4376
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5435781962940966e-06,
+      "loss": 0.509,
+      "step": 4377
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.540316834860136e-06,
+      "loss": 0.5227,
+      "step": 4378
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5370586349509886e-06,
+      "loss": 0.501,
+      "step": 4379
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5338035977842948e-06,
+      "loss": 0.5291,
+      "step": 4380
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5305517245765177e-06,
+      "loss": 0.5498,
+      "step": 4381
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.527303016542937e-06,
+      "loss": 0.4921,
+      "step": 4382
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.524057474897649e-06,
+      "loss": 0.4954,
+      "step": 4383
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5208151008535676e-06,
+      "loss": 0.5256,
+      "step": 4384
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5175758956224229e-06,
+      "loss": 0.5275,
+      "step": 4385
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5143398604147552e-06,
+      "loss": 0.5186,
+      "step": 4386
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5111069964399294e-06,
+      "loss": 0.5731,
+      "step": 4387
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5078773049061214e-06,
+      "loss": 0.5034,
+      "step": 4388
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5046507870203153e-06,
+      "loss": 0.5592,
+      "step": 4389
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5014274439883192e-06,
+      "loss": 0.5128,
+      "step": 4390
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4982072770147517e-06,
+      "loss": 0.5167,
+      "step": 4391
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4949902873030386e-06,
+      "loss": 0.4941,
+      "step": 4392
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4917764760554231e-06,
+      "loss": 0.5277,
+      "step": 4393
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4885658444729622e-06,
+      "loss": 0.5328,
+      "step": 4394
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.485358393755524e-06,
+      "loss": 0.5323,
+      "step": 4395
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4821541251017823e-06,
+      "loss": 0.4895,
+      "step": 4396
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4789530397092278e-06,
+      "loss": 0.5365,
+      "step": 4397
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4757551387741587e-06,
+      "loss": 0.4797,
+      "step": 4398
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4725604234916846e-06,
+      "loss": 0.493,
+      "step": 4399
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.469368895055725e-06,
+      "loss": 0.5422,
+      "step": 4400
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.466180554659008e-06,
+      "loss": 0.5018,
+      "step": 4401
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.462995403493066e-06,
+      "loss": 0.505,
+      "step": 4402
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4598134427482446e-06,
+      "loss": 0.5375,
+      "step": 4403
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4566346736136994e-06,
+      "loss": 0.5514,
+      "step": 4404
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4534590972773833e-06,
+      "loss": 0.4822,
+      "step": 4405
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4502867149260658e-06,
+      "loss": 0.5134,
+      "step": 4406
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4471175277453198e-06,
+      "loss": 0.5193,
+      "step": 4407
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4439515369195201e-06,
+      "loss": 0.5129,
+      "step": 4408
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4407887436318535e-06,
+      "loss": 0.5071,
+      "step": 4409
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.437629149064307e-06,
+      "loss": 0.4835,
+      "step": 4410
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4344727543976754e-06,
+      "loss": 0.5341,
+      "step": 4411
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4313195608115537e-06,
+      "loss": 0.4987,
+      "step": 4412
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4281695694843444e-06,
+      "loss": 0.527,
+      "step": 4413
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4250227815932516e-06,
+      "loss": 0.4924,
+      "step": 4414
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4218791983142831e-06,
+      "loss": 0.515,
+      "step": 4415
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4187388208222474e-06,
+      "loss": 0.5675,
+      "step": 4416
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4156016502907588e-06,
+      "loss": 0.553,
+      "step": 4417
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4124676878922262e-06,
+      "loss": 0.5201,
+      "step": 4418
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4093369347978658e-06,
+      "loss": 0.5466,
+      "step": 4419
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.406209392177693e-06,
+      "loss": 0.5479,
+      "step": 4420
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.403085061200521e-06,
+      "loss": 0.506,
+      "step": 4421
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.399963943033964e-06,
+      "loss": 0.5286,
+      "step": 4422
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3968460388444394e-06,
+      "loss": 0.5544,
+      "step": 4423
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.393731349797156e-06,
+      "loss": 0.5056,
+      "step": 4424
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3906198770561264e-06,
+      "loss": 0.528,
+      "step": 4425
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.38751162178416e-06,
+      "loss": 0.5466,
+      "step": 4426
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3844065851428645e-06,
+      "loss": 0.5065,
+      "step": 4427
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.381304768292644e-06,
+      "loss": 0.5159,
+      "step": 4428
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3782061723926975e-06,
+      "loss": 0.5385,
+      "step": 4429
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.375110798601026e-06,
+      "loss": 0.5354,
+      "step": 4430
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.372018648074417e-06,
+      "loss": 0.5349,
+      "step": 4431
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3689297219684615e-06,
+      "loss": 0.5084,
+      "step": 4432
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3658440214375446e-06,
+      "loss": 0.528,
+      "step": 4433
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3627615476348398e-06,
+      "loss": 0.5016,
+      "step": 4434
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3596823017123218e-06,
+      "loss": 0.5233,
+      "step": 4435
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3566062848207573e-06,
+      "loss": 0.5271,
+      "step": 4436
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3535334981097016e-06,
+      "loss": 0.5356,
+      "step": 4437
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.350463942727508e-06,
+      "loss": 0.5353,
+      "step": 4438
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3473976198213223e-06,
+      "loss": 0.5597,
+      "step": 4439
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.344334530537078e-06,
+      "loss": 0.5253,
+      "step": 4440
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3412746760195038e-06,
+      "loss": 0.5436,
+      "step": 4441
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3382180574121185e-06,
+      "loss": 0.5212,
+      "step": 4442
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3351646758572302e-06,
+      "loss": 0.5044,
+      "step": 4443
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3321145324959407e-06,
+      "loss": 0.527,
+      "step": 4444
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3290676284681392e-06,
+      "loss": 0.5383,
+      "step": 4445
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.326023964912505e-06,
+      "loss": 0.5186,
+      "step": 4446
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3229835429665039e-06,
+      "loss": 0.5018,
+      "step": 4447
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3199463637663922e-06,
+      "loss": 0.5192,
+      "step": 4448
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.316912428447219e-06,
+      "loss": 0.5282,
+      "step": 4449
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3138817381428105e-06,
+      "loss": 0.5494,
+      "step": 4450
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3108542939857883e-06,
+      "loss": 0.5753,
+      "step": 4451
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.307830097107562e-06,
+      "loss": 0.5437,
+      "step": 4452
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3048091486383197e-06,
+      "loss": 0.556,
+      "step": 4453
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3017914497070427e-06,
+      "loss": 0.5377,
+      "step": 4454
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2987770014414957e-06,
+      "loss": 0.5029,
+      "step": 4455
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2957658049682275e-06,
+      "loss": 0.4772,
+      "step": 4456
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2927578614125735e-06,
+      "loss": 0.5422,
+      "step": 4457
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2897531718986511e-06,
+      "loss": 0.5254,
+      "step": 4458
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2867517375493655e-06,
+      "loss": 0.498,
+      "step": 4459
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2837535594863992e-06,
+      "loss": 0.5326,
+      "step": 4460
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.280758638830224e-06,
+      "loss": 0.4902,
+      "step": 4461
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2777669767000921e-06,
+      "loss": 0.5186,
+      "step": 4462
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2747785742140338e-06,
+      "loss": 0.5416,
+      "step": 4463
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2717934324888682e-06,
+      "loss": 0.4933,
+      "step": 4464
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2688115526401945e-06,
+      "loss": 0.5323,
+      "step": 4465
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.265832935782385e-06,
+      "loss": 0.5094,
+      "step": 4466
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2628575830286039e-06,
+      "loss": 0.4594,
+      "step": 4467
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.259885495490789e-06,
+      "loss": 0.5309,
+      "step": 4468
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2569166742796557e-06,
+      "loss": 0.53,
+      "step": 4469
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2539511205047062e-06,
+      "loss": 0.5492,
+      "step": 4470
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.250988835274215e-06,
+      "loss": 0.5225,
+      "step": 4471
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2480298196952379e-06,
+      "loss": 0.5021,
+      "step": 4472
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.245074074873609e-06,
+      "loss": 0.5273,
+      "step": 4473
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2421216019139403e-06,
+      "loss": 0.5065,
+      "step": 4474
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2391724019196206e-06,
+      "loss": 0.5021,
+      "step": 4475
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2362264759928121e-06,
+      "loss": 0.5001,
+      "step": 4476
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2332838252344591e-06,
+      "loss": 0.4985,
+      "step": 4477
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2303444507442796e-06,
+      "loss": 0.5577,
+      "step": 4478
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2274083536207638e-06,
+      "loss": 0.5748,
+      "step": 4479
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2244755349611826e-06,
+      "loss": 0.5483,
+      "step": 4480
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2215459958615805e-06,
+      "loss": 0.5638,
+      "step": 4481
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.218619737416772e-06,
+      "loss": 0.4923,
+      "step": 4482
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2156967607203496e-06,
+      "loss": 0.4877,
+      "step": 4483
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.212777066864682e-06,
+      "loss": 0.5565,
+      "step": 4484
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2098606569409023e-06,
+      "loss": 0.5215,
+      "step": 4485
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2069475320389234e-06,
+      "loss": 0.5073,
+      "step": 4486
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2040376932474307e-06,
+      "loss": 0.527,
+      "step": 4487
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2011311416538773e-06,
+      "loss": 0.5233,
+      "step": 4488
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1982278783444901e-06,
+      "loss": 0.4967,
+      "step": 4489
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1953279044042687e-06,
+      "loss": 0.498,
+      "step": 4490
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.192431220916983e-06,
+      "loss": 0.5283,
+      "step": 4491
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1895378289651672e-06,
+      "loss": 0.5537,
+      "step": 4492
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1866477296301337e-06,
+      "loss": 0.5131,
+      "step": 4493
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1837609239919612e-06,
+      "loss": 0.5095,
+      "step": 4494
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1808774131294943e-06,
+      "loss": 0.5112,
+      "step": 4495
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1779971981203509e-06,
+      "loss": 0.5155,
+      "step": 4496
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.175120280040918e-06,
+      "loss": 0.5099,
+      "step": 4497
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1722466599663417e-06,
+      "loss": 0.5636,
+      "step": 4498
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.169376338970546e-06,
+      "loss": 0.5133,
+      "step": 4499
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1665093181262178e-06,
+      "loss": 0.5201,
+      "step": 4500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.163645598504809e-06,
+      "loss": 0.5525,
+      "step": 4501
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1607851811765402e-06,
+      "loss": 0.5234,
+      "step": 4502
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1579280672103998e-06,
+      "loss": 0.527,
+      "step": 4503
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1550742576741347e-06,
+      "loss": 0.4819,
+      "step": 4504
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.152223753634263e-06,
+      "loss": 0.5248,
+      "step": 4505
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1493765561560644e-06,
+      "loss": 0.5167,
+      "step": 4506
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1465326663035881e-06,
+      "loss": 0.5421,
+      "step": 4507
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1436920851396382e-06,
+      "loss": 0.519,
+      "step": 4508
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1408548137257902e-06,
+      "loss": 0.5165,
+      "step": 4509
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1380208531223803e-06,
+      "loss": 0.5254,
+      "step": 4510
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1351902043885033e-06,
+      "loss": 0.5221,
+      "step": 4511
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1323628685820232e-06,
+      "loss": 0.5168,
+      "step": 4512
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1295388467595624e-06,
+      "loss": 0.5422,
+      "step": 4513
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1267181399765036e-06,
+      "loss": 0.4861,
+      "step": 4514
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1239007492869913e-06,
+      "loss": 0.5287,
+      "step": 4515
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1210866757439332e-06,
+      "loss": 0.5046,
+      "step": 4516
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1182759203989958e-06,
+      "loss": 0.4763,
+      "step": 4517
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1154684843026042e-06,
+      "loss": 0.5514,
+      "step": 4518
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1126643685039461e-06,
+      "loss": 0.5483,
+      "step": 4519
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1098635740509634e-06,
+      "loss": 0.5061,
+      "step": 4520
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1070661019903606e-06,
+      "loss": 0.525,
+      "step": 4521
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1042719533676006e-06,
+      "loss": 0.4659,
+      "step": 4522
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1014811292269045e-06,
+      "loss": 0.5453,
+      "step": 4523
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0986936306112473e-06,
+      "loss": 0.5417,
+      "step": 4524
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0959094585623642e-06,
+      "loss": 0.5197,
+      "step": 4525
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0931286141207498e-06,
+      "loss": 0.5325,
+      "step": 4526
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0903510983256482e-06,
+      "loss": 0.5504,
+      "step": 4527
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0875769122150636e-06,
+      "loss": 0.4834,
+      "step": 4528
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.08480605682576e-06,
+      "loss": 0.5154,
+      "step": 4529
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.082038533193247e-06,
+      "loss": 0.5062,
+      "step": 4530
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.079274342351797e-06,
+      "loss": 0.5011,
+      "step": 4531
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.076513485334434e-06,
+      "loss": 0.5363,
+      "step": 4532
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0737559631729356e-06,
+      "loss": 0.5178,
+      "step": 4533
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0710017768978343e-06,
+      "loss": 0.4808,
+      "step": 4534
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.068250927538419e-06,
+      "loss": 0.5475,
+      "step": 4535
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0655034161227218e-06,
+      "loss": 0.505,
+      "step": 4536
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0627592436775369e-06,
+      "loss": 0.5431,
+      "step": 4537
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0600184112284073e-06,
+      "loss": 0.5481,
+      "step": 4538
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0572809197996291e-06,
+      "loss": 0.5048,
+      "step": 4539
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0545467704142453e-06,
+      "loss": 0.5253,
+      "step": 4540
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.051815964094054e-06,
+      "loss": 0.527,
+      "step": 4541
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0490885018596075e-06,
+      "loss": 0.5093,
+      "step": 4542
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0463643847301975e-06,
+      "loss": 0.5035,
+      "step": 4543
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0436436137238748e-06,
+      "loss": 0.5192,
+      "step": 4544
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0409261898574374e-06,
+      "loss": 0.5116,
+      "step": 4545
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0382121141464308e-06,
+      "loss": 0.4858,
+      "step": 4546
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0355013876051522e-06,
+      "loss": 0.5341,
+      "step": 4547
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0327940112466461e-06,
+      "loss": 0.5123,
+      "step": 4548
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0300899860827007e-06,
+      "loss": 0.5458,
+      "step": 4549
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0273893131238578e-06,
+      "loss": 0.5127,
+      "step": 4550
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0246919933794042e-06,
+      "loss": 0.4972,
+      "step": 4551
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0219980278573704e-06,
+      "loss": 0.5436,
+      "step": 4552
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.019307417564539e-06,
+      "loss": 0.5084,
+      "step": 4553
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0166201635064354e-06,
+      "loss": 0.5026,
+      "step": 4554
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0139362666873331e-06,
+      "loss": 0.5079,
+      "step": 4555
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0112557281102452e-06,
+      "loss": 0.5439,
+      "step": 4556
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0085785487769362e-06,
+      "loss": 0.4923,
+      "step": 4557
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0059047296879132e-06,
+      "loss": 0.4917,
+      "step": 4558
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0032342718424248e-06,
+      "loss": 0.5457,
+      "step": 4559
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0005671762384672e-06,
+      "loss": 0.5284,
+      "step": 4560
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.979034438727786e-07,
+      "loss": 0.5326,
+      "step": 4561
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.952430757408393e-07,
+      "loss": 0.5085,
+      "step": 4562
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.925860728368742e-07,
+      "loss": 0.5454,
+      "step": 4563
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.89932436153852e-07,
+      "loss": 0.5645,
+      "step": 4564
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.872821666834753e-07,
+      "loss": 0.5211,
+      "step": 4565
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.846352654161983e-07,
+      "loss": 0.5231,
+      "step": 4566
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.81991733341212e-07,
+      "loss": 0.5488,
+      "step": 4567
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.793515714464464e-07,
+      "loss": 0.5116,
+      "step": 4568
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.76714780718574e-07,
+      "loss": 0.493,
+      "step": 4569
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.740813621430101e-07,
+      "loss": 0.5343,
+      "step": 4570
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.714513167039063e-07,
+      "loss": 0.4877,
+      "step": 4571
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.688246453841532e-07,
+      "loss": 0.5049,
+      "step": 4572
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.662013491653822e-07,
+      "loss": 0.5078,
+      "step": 4573
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.635814290279644e-07,
+      "loss": 0.5264,
+      "step": 4574
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.609648859510056e-07,
+      "loss": 0.4998,
+      "step": 4575
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.583517209123517e-07,
+      "loss": 0.5145,
+      "step": 4576
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.557419348885876e-07,
+      "loss": 0.5085,
+      "step": 4577
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.531355288550326e-07,
+      "loss": 0.5328,
+      "step": 4578
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.505325037857449e-07,
+      "loss": 0.4661,
+      "step": 4579
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.479328606535198e-07,
+      "loss": 0.5137,
+      "step": 4580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.453366004298848e-07,
+      "loss": 0.5204,
+      "step": 4581
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.427437240851056e-07,
+      "loss": 0.5022,
+      "step": 4582
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.401542325881863e-07,
+      "loss": 0.5226,
+      "step": 4583
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.375681269068604e-07,
+      "loss": 0.4993,
+      "step": 4584
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.349854080075982e-07,
+      "loss": 0.4963,
+      "step": 4585
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.324060768556075e-07,
+      "loss": 0.5138,
+      "step": 4586
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.298301344148275e-07,
+      "loss": 0.5117,
+      "step": 4587
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.272575816479279e-07,
+      "loss": 0.5673,
+      "step": 4588
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.246884195163153e-07,
+      "loss": 0.5051,
+      "step": 4589
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.221226489801304e-07,
+      "loss": 0.4916,
+      "step": 4590
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.195602709982432e-07,
+      "loss": 0.5355,
+      "step": 4591
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.170012865282574e-07,
+      "loss": 0.5254,
+      "step": 4592
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.144456965265103e-07,
+      "loss": 0.5524,
+      "step": 4593
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.118935019480645e-07,
+      "loss": 0.5294,
+      "step": 4594
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.093447037467196e-07,
+      "loss": 0.5212,
+      "step": 4595
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.067993028750066e-07,
+      "loss": 0.5134,
+      "step": 4596
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.042573002841792e-07,
+      "loss": 0.5263,
+      "step": 4597
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.017186969242286e-07,
+      "loss": 0.5161,
+      "step": 4598
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.991834937438759e-07,
+      "loss": 0.5126,
+      "step": 4599
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.966516916905643e-07,
+      "loss": 0.5344,
+      "step": 4600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.941232917104736e-07,
+      "loss": 0.5552,
+      "step": 4601
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.915982947485069e-07,
+      "loss": 0.5303,
+      "step": 4602
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.890767017483015e-07,
+      "loss": 0.5426,
+      "step": 4603
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.865585136522136e-07,
+      "loss": 0.5098,
+      "step": 4604
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.840437314013351e-07,
+      "loss": 0.5388,
+      "step": 4605
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.815323559354816e-07,
+      "loss": 0.4988,
+      "step": 4606
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.79024388193197e-07,
+      "loss": 0.5327,
+      "step": 4607
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.765198291117482e-07,
+      "loss": 0.5393,
+      "step": 4608
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.740186796271355e-07,
+      "loss": 0.5284,
+      "step": 4609
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.715209406740755e-07,
+      "loss": 0.515,
+      "step": 4610
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.690266131860159e-07,
+      "loss": 0.5688,
+      "step": 4611
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.665356980951312e-07,
+      "loss": 0.5199,
+      "step": 4612
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.640481963323145e-07,
+      "loss": 0.5564,
+      "step": 4613
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.615641088271876e-07,
+      "loss": 0.5284,
+      "step": 4614
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.59083436508098e-07,
+      "loss": 0.5209,
+      "step": 4615
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.566061803021109e-07,
+      "loss": 0.5349,
+      "step": 4616
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.541323411350189e-07,
+      "loss": 0.5326,
+      "step": 4617
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.516619199313381e-07,
+      "loss": 0.5282,
+      "step": 4618
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.491949176143088e-07,
+      "loss": 0.5204,
+      "step": 4619
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.467313351058848e-07,
+      "loss": 0.5747,
+      "step": 4620
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.442711733267527e-07,
+      "loss": 0.5656,
+      "step": 4621
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.41814433196314e-07,
+      "loss": 0.5253,
+      "step": 4622
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.393611156326931e-07,
+      "loss": 0.4906,
+      "step": 4623
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.369112215527375e-07,
+      "loss": 0.5099,
+      "step": 4624
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.344647518720151e-07,
+      "loss": 0.5238,
+      "step": 4625
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.320217075048076e-07,
+      "loss": 0.5227,
+      "step": 4626
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.295820893641238e-07,
+      "loss": 0.5596,
+      "step": 4627
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.27145898361692e-07,
+      "loss": 0.5293,
+      "step": 4628
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.247131354079541e-07,
+      "loss": 0.5298,
+      "step": 4629
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.222838014120749e-07,
+      "loss": 0.5205,
+      "step": 4630
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.198578972819382e-07,
+      "loss": 0.5322,
+      "step": 4631
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.174354239241445e-07,
+      "loss": 0.5074,
+      "step": 4632
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.150163822440105e-07,
+      "loss": 0.5297,
+      "step": 4633
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.126007731455754e-07,
+      "loss": 0.5187,
+      "step": 4634
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.101885975315915e-07,
+      "loss": 0.5103,
+      "step": 4635
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.077798563035289e-07,
+      "loss": 0.5129,
+      "step": 4636
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.05374550361574e-07,
+      "loss": 0.5128,
+      "step": 4637
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.029726806046323e-07,
+      "loss": 0.5355,
+      "step": 4638
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.005742479303191e-07,
+      "loss": 0.5264,
+      "step": 4639
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.981792532349708e-07,
+      "loss": 0.5315,
+      "step": 4640
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.957876974136381e-07,
+      "loss": 0.5287,
+      "step": 4641
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.933995813600814e-07,
+      "loss": 0.5636,
+      "step": 4642
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.910149059667816e-07,
+      "loss": 0.5066,
+      "step": 4643
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.886336721249343e-07,
+      "loss": 0.526,
+      "step": 4644
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.862558807244414e-07,
+      "loss": 0.5299,
+      "step": 4645
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.838815326539262e-07,
+      "loss": 0.5314,
+      "step": 4646
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.815106288007235e-07,
+      "loss": 0.5774,
+      "step": 4647
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.791431700508767e-07,
+      "loss": 0.5348,
+      "step": 4648
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.767791572891448e-07,
+      "loss": 0.55,
+      "step": 4649
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.74418591399001e-07,
+      "loss": 0.5135,
+      "step": 4650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.720614732626264e-07,
+      "loss": 0.5253,
+      "step": 4651
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.697078037609162e-07,
+      "loss": 0.5187,
+      "step": 4652
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.673575837734759e-07,
+      "loss": 0.5178,
+      "step": 4653
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.650108141786228e-07,
+      "loss": 0.5355,
+      "step": 4654
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.626674958533819e-07,
+      "loss": 0.5399,
+      "step": 4655
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.603276296734918e-07,
+      "loss": 0.5017,
+      "step": 4656
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.579912165134006e-07,
+      "loss": 0.5186,
+      "step": 4657
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.556582572462612e-07,
+      "loss": 0.5138,
+      "step": 4658
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.533287527439425e-07,
+      "loss": 0.5295,
+      "step": 4659
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.510027038770196e-07,
+      "loss": 0.5566,
+      "step": 4660
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.486801115147735e-07,
+      "loss": 0.5234,
+      "step": 4661
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.463609765251955e-07,
+      "loss": 0.5217,
+      "step": 4662
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.440452997749892e-07,
+      "loss": 0.5182,
+      "step": 4663
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.417330821295577e-07,
+      "loss": 0.5066,
+      "step": 4664
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.394243244530152e-07,
+      "loss": 0.5309,
+      "step": 4665
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.371190276081852e-07,
+      "loss": 0.4708,
+      "step": 4666
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.348171924565939e-07,
+      "loss": 0.5247,
+      "step": 4667
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.325188198584765e-07,
+      "loss": 0.5065,
+      "step": 4668
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.302239106727727e-07,
+      "loss": 0.5106,
+      "step": 4669
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.279324657571308e-07,
+      "loss": 0.5295,
+      "step": 4670
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.25644485967899e-07,
+      "loss": 0.5402,
+      "step": 4671
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.233599721601337e-07,
+      "loss": 0.5236,
+      "step": 4672
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.210789251875971e-07,
+      "loss": 0.4965,
+      "step": 4673
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.188013459027543e-07,
+      "loss": 0.5313,
+      "step": 4674
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.165272351567731e-07,
+      "loss": 0.4766,
+      "step": 4675
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.142565937995294e-07,
+      "loss": 0.4934,
+      "step": 4676
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.119894226795976e-07,
+      "loss": 0.5067,
+      "step": 4677
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.097257226442578e-07,
+      "loss": 0.5521,
+      "step": 4678
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.074654945394932e-07,
+      "loss": 0.5265,
+      "step": 4679
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.052087392099871e-07,
+      "loss": 0.5088,
+      "step": 4680
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.029554574991293e-07,
+      "loss": 0.5183,
+      "step": 4681
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.007056502490062e-07,
+      "loss": 0.5228,
+      "step": 4682
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.984593183004118e-07,
+      "loss": 0.5514,
+      "step": 4683
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.962164624928347e-07,
+      "loss": 0.4728,
+      "step": 4684
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.939770836644666e-07,
+      "loss": 0.5177,
+      "step": 4685
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.917411826522058e-07,
+      "loss": 0.5013,
+      "step": 4686
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.895087602916406e-07,
+      "loss": 0.5522,
+      "step": 4687
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.872798174170658e-07,
+      "loss": 0.5167,
+      "step": 4688
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.850543548614774e-07,
+      "loss": 0.5378,
+      "step": 4689
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.828323734565645e-07,
+      "loss": 0.517,
+      "step": 4690
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.806138740327195e-07,
+      "loss": 0.4912,
+      "step": 4691
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.783988574190347e-07,
+      "loss": 0.5305,
+      "step": 4692
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.761873244432949e-07,
+      "loss": 0.5249,
+      "step": 4693
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.739792759319896e-07,
+      "loss": 0.5293,
+      "step": 4694
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.717747127103035e-07,
+      "loss": 0.4758,
+      "step": 4695
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.695736356021176e-07,
+      "loss": 0.5112,
+      "step": 4696
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.673760454300115e-07,
+      "loss": 0.5403,
+      "step": 4697
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.651819430152618e-07,
+      "loss": 0.4882,
+      "step": 4698
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.629913291778434e-07,
+      "loss": 0.5272,
+      "step": 4699
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.608042047364193e-07,
+      "loss": 0.5304,
+      "step": 4700
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.586205705083592e-07,
+      "loss": 0.5475,
+      "step": 4701
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.564404273097236e-07,
+      "loss": 0.5457,
+      "step": 4702
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.54263775955265e-07,
+      "loss": 0.4962,
+      "step": 4703
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.520906172584374e-07,
+      "loss": 0.5428,
+      "step": 4704
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.499209520313865e-07,
+      "loss": 0.5247,
+      "step": 4705
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.477547810849505e-07,
+      "loss": 0.5051,
+      "step": 4706
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.45592105228664e-07,
+      "loss": 0.5607,
+      "step": 4707
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.434329252707572e-07,
+      "loss": 0.4542,
+      "step": 4708
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.41277242018149e-07,
+      "loss": 0.5198,
+      "step": 4709
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.39125056276455e-07,
+      "loss": 0.5411,
+      "step": 4710
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.36976368849983e-07,
+      "loss": 0.521,
+      "step": 4711
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.34831180541734e-07,
+      "loss": 0.5234,
+      "step": 4712
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.32689492153401e-07,
+      "loss": 0.5153,
+      "step": 4713
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.305513044853673e-07,
+      "loss": 0.5266,
+      "step": 4714
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.284166183367113e-07,
+      "loss": 0.5162,
+      "step": 4715
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.262854345051994e-07,
+      "loss": 0.486,
+      "step": 4716
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.241577537872901e-07,
+      "loss": 0.5448,
+      "step": 4717
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.22033576978135e-07,
+      "loss": 0.5379,
+      "step": 4718
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.199129048715724e-07,
+      "loss": 0.5502,
+      "step": 4719
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.17795738260134e-07,
+      "loss": 0.4902,
+      "step": 4720
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.156820779350403e-07,
+      "loss": 0.5488,
+      "step": 4721
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.135719246862016e-07,
+      "loss": 0.5352,
+      "step": 4722
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.114652793022158e-07,
+      "loss": 0.5215,
+      "step": 4723
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.093621425703722e-07,
+      "loss": 0.5253,
+      "step": 4724
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.072625152766498e-07,
+      "loss": 0.5382,
+      "step": 4725
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.051663982057121e-07,
+      "loss": 0.5085,
+      "step": 4726
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.5287,
+      "step": 4727
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.009846978643008e-07,
+      "loss": 0.5713,
+      "step": 4728
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.98899116156596e-07,
+      "loss": 0.5173,
+      "step": 4729
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.968170477972179e-07,
+      "loss": 0.5359,
+      "step": 4730
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.947384935642742e-07,
+      "loss": 0.5179,
+      "step": 4731
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.926634542345499e-07,
+      "loss": 0.5044,
+      "step": 4732
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.905919305835239e-07,
+      "loss": 0.5198,
+      "step": 4733
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.885239233853613e-07,
+      "loss": 0.5208,
+      "step": 4734
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.86459433412907e-07,
+      "loss": 0.5342,
+      "step": 4735
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.843984614376974e-07,
+      "loss": 0.5126,
+      "step": 4736
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.823410082299541e-07,
+      "loss": 0.4791,
+      "step": 4737
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.802870745585787e-07,
+      "loss": 0.5454,
+      "step": 4738
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.782366611911605e-07,
+      "loss": 0.5493,
+      "step": 4739
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.761897688939744e-07,
+      "loss": 0.547,
+      "step": 4740
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.741463984319784e-07,
+      "loss": 0.547,
+      "step": 4741
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.721065505688139e-07,
+      "loss": 0.5038,
+      "step": 4742
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.700702260668079e-07,
+      "loss": 0.4948,
+      "step": 4743
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.680374256869647e-07,
+      "loss": 0.5118,
+      "step": 4744
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.66008150188978e-07,
+      "loss": 0.5291,
+      "step": 4745
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.639824003312222e-07,
+      "loss": 0.5223,
+      "step": 4746
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.61960176870755e-07,
+      "loss": 0.5494,
+      "step": 4747
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.599414805633119e-07,
+      "loss": 0.5054,
+      "step": 4748
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.57926312163316e-07,
+      "loss": 0.5596,
+      "step": 4749
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.559146724238695e-07,
+      "loss": 0.5363,
+      "step": 4750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.539065620967521e-07,
+      "loss": 0.5767,
+      "step": 4751
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.519019819324322e-07,
+      "loss": 0.5397,
+      "step": 4752
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.499009326800542e-07,
+      "loss": 0.5359,
+      "step": 4753
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.479034150874418e-07,
+      "loss": 0.5248,
+      "step": 4754
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.459094299011025e-07,
+      "loss": 0.5316,
+      "step": 4755
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.439189778662202e-07,
+      "loss": 0.5261,
+      "step": 4756
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.419320597266609e-07,
+      "loss": 0.5211,
+      "step": 4757
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.399486762249695e-07,
+      "loss": 0.5144,
+      "step": 4758
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.379688281023698e-07,
+      "loss": 0.5316,
+      "step": 4759
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.359925160987622e-07,
+      "loss": 0.5382,
+      "step": 4760
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.340197409527292e-07,
+      "loss": 0.5424,
+      "step": 4761
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.320505034015288e-07,
+      "loss": 0.5023,
+      "step": 4762
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.300848041811001e-07,
+      "loss": 0.4974,
+      "step": 4763
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.281226440260556e-07,
+      "loss": 0.5442,
+      "step": 4764
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.261640236696864e-07,
+      "loss": 0.5419,
+      "step": 4765
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.242089438439657e-07,
+      "loss": 0.53,
+      "step": 4766
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.222574052795337e-07,
+      "loss": 0.4893,
+      "step": 4767
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.20309408705717e-07,
+      "loss": 0.5493,
+      "step": 4768
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.183649548505143e-07,
+      "loss": 0.5088,
+      "step": 4769
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.164240444405977e-07,
+      "loss": 0.5435,
+      "step": 4770
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.144866782013214e-07,
+      "loss": 0.5507,
+      "step": 4771
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.125528568567107e-07,
+      "loss": 0.5299,
+      "step": 4772
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.10622581129464e-07,
+      "loss": 0.5233,
+      "step": 4773
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.086958517409602e-07,
+      "loss": 0.5222,
+      "step": 4774
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.067726694112507e-07,
+      "loss": 0.515,
+      "step": 4775
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.048530348590586e-07,
+      "loss": 0.5242,
+      "step": 4776
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.029369488017843e-07,
+      "loss": 0.5514,
+      "step": 4777
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.010244119555019e-07,
+      "loss": 0.4523,
+      "step": 4778
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.991154250349594e-07,
+      "loss": 0.5062,
+      "step": 4779
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.97209988753573e-07,
+      "loss": 0.5248,
+      "step": 4780
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.953081038234397e-07,
+      "loss": 0.5342,
+      "step": 4781
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.934097709553254e-07,
+      "loss": 0.5248,
+      "step": 4782
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.915149908586658e-07,
+      "loss": 0.5037,
+      "step": 4783
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.896237642415746e-07,
+      "loss": 0.5405,
+      "step": 4784
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.877360918108342e-07,
+      "loss": 0.4932,
+      "step": 4785
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.858519742718992e-07,
+      "loss": 0.5109,
+      "step": 4786
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.839714123288963e-07,
+      "loss": 0.5228,
+      "step": 4787
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.820944066846223e-07,
+      "loss": 0.5078,
+      "step": 4788
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.80220958040546e-07,
+      "loss": 0.4943,
+      "step": 4789
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.78351067096805e-07,
+      "loss": 0.4836,
+      "step": 4790
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7648473455221054e-07,
+      "loss": 0.5475,
+      "step": 4791
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.746219611042413e-07,
+      "loss": 0.495,
+      "step": 4792
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7276274744904596e-07,
+      "loss": 0.5118,
+      "step": 4793
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.709070942814453e-07,
+      "loss": 0.5114,
+      "step": 4794
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6905500229492916e-07,
+      "loss": 0.521,
+      "step": 4795
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.672064721816516e-07,
+      "loss": 0.4994,
+      "step": 4796
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.653615046324411e-07,
+      "loss": 0.5292,
+      "step": 4797
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6352010033679396e-07,
+      "loss": 0.5459,
+      "step": 4798
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6168225998287075e-07,
+      "loss": 0.5172,
+      "step": 4799
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5984798425750434e-07,
+      "loss": 0.5322,
+      "step": 4800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.580172738461963e-07,
+      "loss": 0.4689,
+      "step": 4801
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.561901294331106e-07,
+      "loss": 0.526,
+      "step": 4802
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5436655170108313e-07,
+      "loss": 0.5278,
+      "step": 4803
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.525465413316177e-07,
+      "loss": 0.5149,
+      "step": 4804
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5073009900487685e-07,
+      "loss": 0.498,
+      "step": 4805
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4891722539969983e-07,
+      "loss": 0.5239,
+      "step": 4806
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.47107921193588e-07,
+      "loss": 0.5321,
+      "step": 4807
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.453021870627061e-07,
+      "loss": 0.5249,
+      "step": 4808
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.435000236818876e-07,
+      "loss": 0.4813,
+      "step": 4809
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4170143172463043e-07,
+      "loss": 0.4773,
+      "step": 4810
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3990641186310246e-07,
+      "loss": 0.5291,
+      "step": 4811
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.381149647681282e-07,
+      "loss": 0.5333,
+      "step": 4812
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.363270911092021e-07,
+      "loss": 0.5422,
+      "step": 4813
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.345427915544831e-07,
+      "loss": 0.517,
+      "step": 4814
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3276206677079437e-07,
+      "loss": 0.5505,
+      "step": 4815
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3098491742362135e-07,
+      "loss": 0.5083,
+      "step": 4816
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.292113441771162e-07,
+      "loss": 0.5173,
+      "step": 4817
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2744134769409086e-07,
+      "loss": 0.4979,
+      "step": 4818
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.256749286360229e-07,
+      "loss": 0.5166,
+      "step": 4819
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2391208766305426e-07,
+      "loss": 0.5419,
+      "step": 4820
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.221528254339846e-07,
+      "loss": 0.4572,
+      "step": 4821
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.203971426062825e-07,
+      "loss": 0.5565,
+      "step": 4822
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1864503983607643e-07,
+      "loss": 0.5219,
+      "step": 4823
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1689651777815477e-07,
+      "loss": 0.5514,
+      "step": 4824
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.151515770859682e-07,
+      "loss": 0.5491,
+      "step": 4825
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1341021841163286e-07,
+      "loss": 0.5042,
+      "step": 4826
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1167244240592376e-07,
+      "loss": 0.4871,
+      "step": 4827
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0993824971827354e-07,
+      "loss": 0.5455,
+      "step": 4828
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.082076409967806e-07,
+      "loss": 0.5283,
+      "step": 4829
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.064806168882041e-07,
+      "loss": 0.527,
+      "step": 4830
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.047571780379589e-07,
+      "loss": 0.4768,
+      "step": 4831
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.030373250901254e-07,
+      "loss": 0.5325,
+      "step": 4832
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0132105868744053e-07,
+      "loss": 0.5132,
+      "step": 4833
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9960837947130127e-07,
+      "loss": 0.5282,
+      "step": 4834
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9789928808176334e-07,
+      "loss": 0.5547,
+      "step": 4835
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.961937851575459e-07,
+      "loss": 0.52,
+      "step": 4836
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.944918713360213e-07,
+      "loss": 0.4919,
+      "step": 4837
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9279354725322295e-07,
+      "loss": 0.4867,
+      "step": 4838
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9109881354384426e-07,
+      "loss": 0.5315,
+      "step": 4839
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8940767084123533e-07,
+      "loss": 0.524,
+      "step": 4840
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8772011977740275e-07,
+      "loss": 0.5288,
+      "step": 4841
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.860361609830143e-07,
+      "loss": 0.5334,
+      "step": 4842
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.843557950873944e-07,
+      "loss": 0.5214,
+      "step": 4843
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8267902271852067e-07,
+      "loss": 0.5079,
+      "step": 4844
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.810058445030329e-07,
+      "loss": 0.5228,
+      "step": 4845
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.793362610662255e-07,
+      "loss": 0.5104,
+      "step": 4846
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.776702730320503e-07,
+      "loss": 0.5255,
+      "step": 4847
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.760078810231138e-07,
+      "loss": 0.5347,
+      "step": 4848
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.743490856606813e-07,
+      "loss": 0.5187,
+      "step": 4849
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.726938875646702e-07,
+      "loss": 0.5428,
+      "step": 4850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7104228735365563e-07,
+      "loss": 0.516,
+      "step": 4851
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.693942856448718e-07,
+      "loss": 0.4591,
+      "step": 4852
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6774988305419925e-07,
+      "loss": 0.5009,
+      "step": 4853
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.66109080196182e-07,
+      "loss": 0.5443,
+      "step": 4854
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6447187768401506e-07,
+      "loss": 0.5271,
+      "step": 4855
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.628382761295468e-07,
+      "loss": 0.5371,
+      "step": 4856
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.612082761432822e-07,
+      "loss": 0.5277,
+      "step": 4857
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5958187833438074e-07,
+      "loss": 0.5365,
+      "step": 4858
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5795908331065274e-07,
+      "loss": 0.5298,
+      "step": 4859
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.563398916785654e-07,
+      "loss": 0.5048,
+      "step": 4860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.547243040432358e-07,
+      "loss": 0.5013,
+      "step": 4861
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5311232100843864e-07,
+      "loss": 0.5179,
+      "step": 4862
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.515039431765954e-07,
+      "loss": 0.5299,
+      "step": 4863
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4989917114878646e-07,
+      "loss": 0.5009,
+      "step": 4864
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.48298005524742e-07,
+      "loss": 0.5321,
+      "step": 4865
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.467004469028412e-07,
+      "loss": 0.5444,
+      "step": 4866
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.45106495880122e-07,
+      "loss": 0.5428,
+      "step": 4867
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4351615305226904e-07,
+      "loss": 0.5048,
+      "step": 4868
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.419294190136191e-07,
+      "loss": 0.5219,
+      "step": 4869
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4034629435716225e-07,
+      "loss": 0.5066,
+      "step": 4870
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3876677967453974e-07,
+      "loss": 0.5223,
+      "step": 4871
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3719087555603935e-07,
+      "loss": 0.5189,
+      "step": 4872
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3561858259060556e-07,
+      "loss": 0.5082,
+      "step": 4873
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3404990136582825e-07,
+      "loss": 0.5092,
+      "step": 4874
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3248483246795196e-07,
+      "loss": 0.4785,
+      "step": 4875
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.309233764818687e-07,
+      "loss": 0.52,
+      "step": 4876
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.293655339911195e-07,
+      "loss": 0.4988,
+      "step": 4877
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2781130557789754e-07,
+      "loss": 0.5001,
+      "step": 4878
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.262606918230427e-07,
+      "loss": 0.4749,
+      "step": 4879
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2471369330604577e-07,
+      "loss": 0.5189,
+      "step": 4880
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2317031060504544e-07,
+      "loss": 0.5004,
+      "step": 4881
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.216305442968304e-07,
+      "loss": 0.5679,
+      "step": 4882
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.200943949568358e-07,
+      "loss": 0.5209,
+      "step": 4883
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1856186315914806e-07,
+      "loss": 0.5466,
+      "step": 4884
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.17032949476499e-07,
+      "loss": 0.4751,
+      "step": 4885
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.155076544802682e-07,
+      "loss": 0.5184,
+      "step": 4886
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.139859787404842e-07,
+      "loss": 0.4865,
+      "step": 4887
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1246792282582316e-07,
+      "loss": 0.5159,
+      "step": 4888
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1095348730360907e-07,
+      "loss": 0.5049,
+      "step": 4889
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0944267273981033e-07,
+      "loss": 0.566,
+      "step": 4890
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.079354796990441e-07,
+      "loss": 0.5028,
+      "step": 4891
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.064319087445722e-07,
+      "loss": 0.5448,
+      "step": 4892
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0493196043830607e-07,
+      "loss": 0.5229,
+      "step": 4893
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0343563534080187e-07,
+      "loss": 0.4844,
+      "step": 4894
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0194293401125894e-07,
+      "loss": 0.5258,
+      "step": 4895
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.004538570075255e-07,
+      "loss": 0.5404,
+      "step": 4896
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9896840488609527e-07,
+      "loss": 0.525,
+      "step": 4897
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9748657820210526e-07,
+      "loss": 0.553,
+      "step": 4898
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.960083775093403e-07,
+      "loss": 0.5326,
+      "step": 4899
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.945338033602285e-07,
+      "loss": 0.5295,
+      "step": 4900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.930628563058413e-07,
+      "loss": 0.5608,
+      "step": 4901
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.915955368958967e-07,
+      "loss": 0.5594,
+      "step": 4902
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.901318456787583e-07,
+      "loss": 0.507,
+      "step": 4903
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8867178320142964e-07,
+      "loss": 0.5518,
+      "step": 4904
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.872153500095609e-07,
+      "loss": 0.5315,
+      "step": 4905
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8576254664744674e-07,
+      "loss": 0.5413,
+      "step": 4906
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8431337365802394e-07,
+      "loss": 0.531,
+      "step": 4907
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8286783158287144e-07,
+      "loss": 0.5396,
+      "step": 4908
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8142592096221166e-07,
+      "loss": 0.4976,
+      "step": 4909
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7998764233491347e-07,
+      "loss": 0.5247,
+      "step": 4910
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7855299623848364e-07,
+      "loss": 0.5106,
+      "step": 4911
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.771219832090732e-07,
+      "loss": 0.5068,
+      "step": 4912
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.756946037814778e-07,
+      "loss": 0.5276,
+      "step": 4913
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.742708584891296e-07,
+      "loss": 0.532,
+      "step": 4914
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7285074786410846e-07,
+      "loss": 0.5432,
+      "step": 4915
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.714342724371333e-07,
+      "loss": 0.5156,
+      "step": 4916
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.700214327375639e-07,
+      "loss": 0.5103,
+      "step": 4917
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6861222929340126e-07,
+      "loss": 0.517,
+      "step": 4918
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6720666263128856e-07,
+      "loss": 0.5226,
+      "step": 4919
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6580473327651123e-07,
+      "loss": 0.5156,
+      "step": 4920
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6440644175299236e-07,
+      "loss": 0.5739,
+      "step": 4921
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.630117885832961e-07,
+      "loss": 0.578,
+      "step": 4922
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6162077428863117e-07,
+      "loss": 0.5616,
+      "step": 4923
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.602333993888384e-07,
+      "loss": 0.5306,
+      "step": 4924
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.588496644024041e-07,
+      "loss": 0.5148,
+      "step": 4925
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5746956984645576e-07,
+      "loss": 0.546,
+      "step": 4926
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5609311623675417e-07,
+      "loss": 0.5375,
+      "step": 4927
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.547203040877044e-07,
+      "loss": 0.5277,
+      "step": 4928
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.533511339123507e-07,
+      "loss": 0.488,
+      "step": 4929
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5198560622237354e-07,
+      "loss": 0.521,
+      "step": 4930
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5062372152809265e-07,
+      "loss": 0.5383,
+      "step": 4931
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4926548033846756e-07,
+      "loss": 0.4869,
+      "step": 4932
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.479108831610966e-07,
+      "loss": 0.5109,
+      "step": 4933
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4655993050221485e-07,
+      "loss": 0.536,
+      "step": 4934
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4521262286669623e-07,
+      "loss": 0.5272,
+      "step": 4935
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4386896075805135e-07,
+      "loss": 0.5544,
+      "step": 4936
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.425289446784296e-07,
+      "loss": 0.5323,
+      "step": 4937
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4119257512861595e-07,
+      "loss": 0.5486,
+      "step": 4938
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3985985260803754e-07,
+      "loss": 0.521,
+      "step": 4939
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3853077761475165e-07,
+      "loss": 0.5275,
+      "step": 4940
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3720535064545635e-07,
+      "loss": 0.4998,
+      "step": 4941
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3588357219548775e-07,
+      "loss": 0.5345,
+      "step": 4942
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.34565442758814e-07,
+      "loss": 0.5808,
+      "step": 4943
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3325096282804326e-07,
+      "loss": 0.5011,
+      "step": 4944
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.319401328944193e-07,
+      "loss": 0.5758,
+      "step": 4945
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3063295344782022e-07,
+      "loss": 0.4999,
+      "step": 4946
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.293294249767597e-07,
+      "loss": 0.5469,
+      "step": 4947
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2802954796839028e-07,
+      "loss": 0.5249,
+      "step": 4948
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.267333229084978e-07,
+      "loss": 0.4998,
+      "step": 4949
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2544075028150037e-07,
+      "loss": 0.5139,
+      "step": 4950
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.241518305704582e-07,
+      "loss": 0.5947,
+      "step": 4951
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.228665642570582e-07,
+      "loss": 0.5487,
+      "step": 4952
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2158495182162732e-07,
+      "loss": 0.5063,
+      "step": 4953
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2030699374312568e-07,
+      "loss": 0.543,
+      "step": 4954
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.19032690499148e-07,
+      "loss": 0.5167,
+      "step": 4955
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1776204256592104e-07,
+      "loss": 0.52,
+      "step": 4956
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.164950504183083e-07,
+      "loss": 0.5408,
+      "step": 4957
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1523171452980773e-07,
+      "loss": 0.5049,
+      "step": 4958
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1397203537254497e-07,
+      "loss": 0.5126,
+      "step": 4959
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1271601341728565e-07,
+      "loss": 0.5692,
+      "step": 4960
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1146364913342655e-07,
+      "loss": 0.5007,
+      "step": 4961
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1021494298899547e-07,
+      "loss": 0.5502,
+      "step": 4962
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0896989545065582e-07,
+      "loss": 0.5331,
+      "step": 4963
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0772850698370317e-07,
+      "loss": 0.5259,
+      "step": 4964
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0649077805206307e-07,
+      "loss": 0.5398,
+      "step": 4965
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0525670911829776e-07,
+      "loss": 0.5166,
+      "step": 4966
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.040263006435983e-07,
+      "loss": 0.5085,
+      "step": 4967
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.027995530877891e-07,
+      "loss": 0.5194,
+      "step": 4968
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0157646690932563e-07,
+      "loss": 0.5303,
+      "step": 4969
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0035704256529565e-07,
+      "loss": 0.4937,
+      "step": 4970
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9914128051142013e-07,
+      "loss": 0.4988,
+      "step": 4971
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9792918120204784e-07,
+      "loss": 0.528,
+      "step": 4972
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.967207450901598e-07,
+      "loss": 0.5223,
+      "step": 4973
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9551597262737255e-07,
+      "loss": 0.4994,
+      "step": 4974
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9431486426392476e-07,
+      "loss": 0.4911,
+      "step": 4975
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.931174204486941e-07,
+      "loss": 0.5283,
+      "step": 4976
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9192364162918363e-07,
+      "loss": 0.4919,
+      "step": 4977
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.907335282515288e-07,
+      "loss": 0.5491,
+      "step": 4978
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.89547080760496e-07,
+      "loss": 0.5184,
+      "step": 4979
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8836429959947945e-07,
+      "loss": 0.5519,
+      "step": 4980
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8718518521050333e-07,
+      "loss": 0.5126,
+      "step": 4981
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8600973803422406e-07,
+      "loss": 0.5409,
+      "step": 4982
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.848379585099258e-07,
+      "loss": 0.5087,
+      "step": 4983
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8366984707552048e-07,
+      "loss": 0.5104,
+      "step": 4984
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8250540416755113e-07,
+      "loss": 0.5356,
+      "step": 4985
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8134463022119076e-07,
+      "loss": 0.5382,
+      "step": 4986
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8018752567024123e-07,
+      "loss": 0.5129,
+      "step": 4987
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7903409094712775e-07,
+      "loss": 0.506,
+      "step": 4988
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7788432648291108e-07,
+      "loss": 0.5061,
+      "step": 4989
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.767382327072775e-07,
+      "loss": 0.4972,
+      "step": 4990
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7559581004853888e-07,
+      "loss": 0.5144,
+      "step": 4991
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7445705893364028e-07,
+      "loss": 0.5663,
+      "step": 4992
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7332197978815134e-07,
+      "loss": 0.4792,
+      "step": 4993
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.721905730362683e-07,
+      "loss": 0.5187,
+      "step": 4994
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7106283910081955e-07,
+      "loss": 0.5543,
+      "step": 4995
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.699387784032569e-07,
+      "loss": 0.5121,
+      "step": 4996
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6881839136365984e-07,
+      "loss": 0.5212,
+      "step": 4997
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6770167840073682e-07,
+      "loss": 0.5204,
+      "step": 4998
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6658863993182174e-07,
+      "loss": 0.5034,
+      "step": 4999
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.654792763728752e-07,
+      "loss": 0.475,
+      "step": 5000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6437358813848557e-07,
+      "loss": 0.5115,
+      "step": 5001
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.632715756418657e-07,
+      "loss": 0.5227,
+      "step": 5002
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6217323929485828e-07,
+      "loss": 0.5149,
+      "step": 5003
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6107857950792838e-07,
+      "loss": 0.4965,
+      "step": 5004
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5998759669016873e-07,
+      "loss": 0.5226,
+      "step": 5005
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5890029124929873e-07,
+      "loss": 0.5251,
+      "step": 5006
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5781666359166004e-07,
+      "loss": 0.5522,
+      "step": 5007
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5673671412222535e-07,
+      "loss": 0.5354,
+      "step": 5008
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5566044324458852e-07,
+      "loss": 0.5115,
+      "step": 5009
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5458785136097e-07,
+      "loss": 0.5292,
+      "step": 5010
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5351893887221582e-07,
+      "loss": 0.5173,
+      "step": 5011
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.524537061777964e-07,
+      "loss": 0.5465,
+      "step": 5012
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5139215367580562e-07,
+      "loss": 0.5275,
+      "step": 5013
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5033428176296383e-07,
+      "loss": 0.486,
+      "step": 5014
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4928009083461704e-07,
+      "loss": 0.5067,
+      "step": 5015
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.482295812847323e-07,
+      "loss": 0.5121,
+      "step": 5016
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4718275350590338e-07,
+      "loss": 0.5127,
+      "step": 5017
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4613960788934734e-07,
+      "loss": 0.5162,
+      "step": 5018
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4510014482490454e-07,
+      "loss": 0.551,
+      "step": 5019
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.440643647010398e-07,
+      "loss": 0.5338,
+      "step": 5020
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4303226790484125e-07,
+      "loss": 0.5283,
+      "step": 5021
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.420038548220215e-07,
+      "loss": 0.5452,
+      "step": 5022
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.409791258369153e-07,
+      "loss": 0.5414,
+      "step": 5023
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3995808133247967e-07,
+      "loss": 0.5522,
+      "step": 5024
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3894072169029937e-07,
+      "loss": 0.5588,
+      "step": 5025
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.379270472905747e-07,
+      "loss": 0.5499,
+      "step": 5026
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.36917058512136e-07,
+      "loss": 0.527,
+      "step": 5027
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3591075573243129e-07,
+      "loss": 0.5029,
+      "step": 5028
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.34908139327532e-07,
+      "loss": 0.5262,
+      "step": 5029
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3390920967213282e-07,
+      "loss": 0.5288,
+      "step": 5030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3291396713955295e-07,
+      "loss": 0.519,
+      "step": 5031
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3192241210172817e-07,
+      "loss": 0.4808,
+      "step": 5032
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3093454492922097e-07,
+      "loss": 0.5385,
+      "step": 5033
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.299503659912127e-07,
+      "loss": 0.5691,
+      "step": 5034
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.28969875655508e-07,
+      "loss": 0.4917,
+      "step": 5035
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2799307428853159e-07,
+      "loss": 0.5278,
+      "step": 5036
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2701996225533253e-07,
+      "loss": 0.5499,
+      "step": 5037
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.260505399195766e-07,
+      "loss": 0.538,
+      "step": 5038
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.25084807643554e-07,
+      "loss": 0.5472,
+      "step": 5039
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2412276578817607e-07,
+      "loss": 0.5358,
+      "step": 5040
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.231644147129729e-07,
+      "loss": 0.4891,
+      "step": 5041
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2220975477609697e-07,
+      "loss": 0.5204,
+      "step": 5042
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2125878633431954e-07,
+      "loss": 0.4817,
+      "step": 5043
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2031150974303407e-07,
+      "loss": 0.504,
+      "step": 5044
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1936792535625296e-07,
+      "loss": 0.5203,
+      "step": 5045
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.184280335266108e-07,
+      "loss": 0.4554,
+      "step": 5046
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1749183460535995e-07,
+      "loss": 0.5042,
+      "step": 5047
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1655932894237276e-07,
+      "loss": 0.5253,
+      "step": 5048
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.156305168861438e-07,
+      "loss": 0.5394,
+      "step": 5049
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1470539878378539e-07,
+      "loss": 0.5176,
+      "step": 5050
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1378397498102878e-07,
+      "loss": 0.5302,
+      "step": 5051
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1286624582222517e-07,
+      "loss": 0.4993,
+      "step": 5052
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1195221165034686e-07,
+      "loss": 0.5082,
+      "step": 5053
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1104187280698176e-07,
+      "loss": 0.5004,
+      "step": 5054
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1013522963233991e-07,
+      "loss": 0.5394,
+      "step": 5055
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0923228246524919e-07,
+      "loss": 0.5226,
+      "step": 5056
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0833303164315523e-07,
+      "loss": 0.5077,
+      "step": 5057
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0743747750212363e-07,
+      "loss": 0.5046,
+      "step": 5058
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.065456203768378e-07,
+      "loss": 0.5676,
+      "step": 5059
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.056574606006e-07,
+      "loss": 0.501,
+      "step": 5060
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0477299850532808e-07,
+      "loss": 0.4908,
+      "step": 5061
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0389223442156315e-07,
+      "loss": 0.5353,
+      "step": 5062
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0301516867846084e-07,
+      "loss": 0.545,
+      "step": 5063
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.021418016037945e-07,
+      "loss": 0.5154,
+      "step": 5064
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0127213352395637e-07,
+      "loss": 0.5258,
+      "step": 5065
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0040616476395537e-07,
+      "loss": 0.5411,
+      "step": 5066
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.954389564741928e-08,
+      "loss": 0.5786,
+      "step": 5067
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.868532649659256e-08,
+      "loss": 0.5044,
+      "step": 5068
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.783045763233634e-08,
+      "loss": 0.5542,
+      "step": 5069
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.697928937412948e-08,
+      "loss": 0.5071,
+      "step": 5070
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.613182204006755e-08,
+      "loss": 0.4614,
+      "step": 5071
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.528805594686274e-08,
+      "loss": 0.5245,
+      "step": 5072
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.444799140984618e-08,
+      "loss": 0.5062,
+      "step": 5073
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.361162874296336e-08,
+      "loss": 0.5046,
+      "step": 5074
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.277896825877541e-08,
+      "loss": 0.5209,
+      "step": 5075
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.195001026846451e-08,
+      "loss": 0.5085,
+      "step": 5076
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.112475508182284e-08,
+      "loss": 0.5077,
+      "step": 5077
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.03032030072637e-08,
+      "loss": 0.5078,
+      "step": 5078
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.948535435181594e-08,
+      "loss": 0.5223,
+      "step": 5079
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.867120942112062e-08,
+      "loss": 0.521,
+      "step": 5080
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.786076851943881e-08,
+      "loss": 0.519,
+      "step": 5081
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.70540319496449e-08,
+      "loss": 0.5282,
+      "step": 5082
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.625100001323106e-08,
+      "loss": 0.5394,
+      "step": 5083
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.545167301030389e-08,
+      "loss": 0.5082,
+      "step": 5084
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.465605123958331e-08,
+      "loss": 0.4768,
+      "step": 5085
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.386413499840929e-08,
+      "loss": 0.5107,
+      "step": 5086
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.307592458273062e-08,
+      "loss": 0.4752,
+      "step": 5087
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.229142028711723e-08,
+      "loss": 0.531,
+      "step": 5088
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.151062240475238e-08,
+      "loss": 0.532,
+      "step": 5089
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.073353122743044e-08,
+      "loss": 0.548,
+      "step": 5090
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.996014704556465e-08,
+      "loss": 0.5447,
+      "step": 5091
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.919047014818159e-08,
+      "loss": 0.509,
+      "step": 5092
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.84245008229223e-08,
+      "loss": 0.5394,
+      "step": 5093
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.766223935604223e-08,
+      "loss": 0.5215,
+      "step": 5094
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.690368603241239e-08,
+      "loss": 0.5082,
+      "step": 5095
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.61488411355149e-08,
+      "loss": 0.4644,
+      "step": 5096
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.539770494744858e-08,
+      "loss": 0.5232,
+      "step": 5097
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.465027774892552e-08,
+      "loss": 0.5038,
+      "step": 5098
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.390655981927342e-08,
+      "loss": 0.5005,
+      "step": 5099
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.316655143642881e-08,
+      "loss": 0.5251,
+      "step": 5100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.243025287694827e-08,
+      "loss": 0.504,
+      "step": 5101
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.169766441599724e-08,
+      "loss": 0.5226,
+      "step": 5102
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.096878632735671e-08,
+      "loss": 0.5132,
+      "step": 5103
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.02436188834199e-08,
+      "loss": 0.5299,
+      "step": 5104
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.95221623551956e-08,
+      "loss": 0.5215,
+      "step": 5105
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.880441701230256e-08,
+      "loss": 0.5134,
+      "step": 5106
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.809038312297511e-08,
+      "loss": 0.5039,
+      "step": 5107
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.738006095405869e-08,
+      "loss": 0.5438,
+      "step": 5108
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.667345077101428e-08,
+      "loss": 0.5559,
+      "step": 5109
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.597055283791176e-08,
+      "loss": 0.5137,
+      "step": 5110
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.527136741743656e-08,
+      "loss": 0.5071,
+      "step": 5111
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.457589477088632e-08,
+      "loss": 0.537,
+      "step": 5112
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.388413515817094e-08,
+      "loss": 0.5356,
+      "step": 5113
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.319608883781247e-08,
+      "loss": 0.5144,
+      "step": 5114
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.251175606694526e-08,
+      "loss": 0.5312,
+      "step": 5115
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.183113710131583e-08,
+      "loss": 0.5117,
+      "step": 5116
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.115423219528293e-08,
+      "loss": 0.5154,
+      "step": 5117
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.048104160181645e-08,
+      "loss": 0.5337,
+      "step": 5118
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.981156557249957e-08,
+      "loss": 0.5226,
+      "step": 5119
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9145804357526595e-08,
+      "loss": 0.5314,
+      "step": 5120
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.848375820570407e-08,
+      "loss": 0.5244,
+      "step": 5121
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.782542736444963e-08,
+      "loss": 0.5211,
+      "step": 5122
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7170812079790916e-08,
+      "loss": 0.53,
+      "step": 5123
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.651991259637113e-08,
+      "loss": 0.5394,
+      "step": 5124
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.587272915744013e-08,
+      "loss": 0.5482,
+      "step": 5125
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.522926200486112e-08,
+      "loss": 0.5261,
+      "step": 5126
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.458951137911061e-08,
+      "loss": 0.5293,
+      "step": 5127
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3953477519272936e-08,
+      "loss": 0.5553,
+      "step": 5128
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.332116066304349e-08,
+      "loss": 0.4989,
+      "step": 5129
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.269256104673215e-08,
+      "loss": 0.5332,
+      "step": 5130
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.206767890525433e-08,
+      "loss": 0.5327,
+      "step": 5131
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1446514472141e-08,
+      "loss": 0.511,
+      "step": 5132
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.082906797953091e-08,
+      "loss": 0.5144,
+      "step": 5133
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.021533965817393e-08,
+      "loss": 0.5427,
+      "step": 5134
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.960532973742993e-08,
+      "loss": 0.5332,
+      "step": 5135
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8999038445270986e-08,
+      "loss": 0.5067,
+      "step": 5136
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8396466008278075e-08,
+      "loss": 0.5345,
+      "step": 5137
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.779761265164218e-08,
+      "loss": 0.5093,
+      "step": 5138
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.720247859916538e-08,
+      "loss": 0.509,
+      "step": 5139
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.661106407325866e-08,
+      "loss": 0.5394,
+      "step": 5140
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6023369294942996e-08,
+      "loss": 0.5358,
+      "step": 5141
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5439394483850486e-08,
+      "loss": 0.5212,
+      "step": 5142
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.485913985822321e-08,
+      "loss": 0.4902,
+      "step": 5143
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4282605634909934e-08,
+      "loss": 0.506,
+      "step": 5144
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.370979202937275e-08,
+      "loss": 0.5295,
+      "step": 5145
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.314069925568043e-08,
+      "loss": 0.514,
+      "step": 5146
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2575327526513944e-08,
+      "loss": 0.5204,
+      "step": 5147
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2013677053160954e-08,
+      "loss": 0.569,
+      "step": 5148
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.1455748045519104e-08,
+      "loss": 0.4923,
+      "step": 5149
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0901540712098286e-08,
+      "loss": 0.5099,
+      "step": 5150
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.035105526001171e-08,
+      "loss": 0.4883,
+      "step": 5151
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.980429189498702e-08,
+      "loss": 0.545,
+      "step": 5152
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9261250821359673e-08,
+      "loss": 0.5333,
+      "step": 5153
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.872193224207066e-08,
+      "loss": 0.545,
+      "step": 5154
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8186336358674305e-08,
+      "loss": 0.5004,
+      "step": 5155
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7654463371330495e-08,
+      "loss": 0.5361,
+      "step": 5156
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.712631347880913e-08,
+      "loss": 0.5491,
+      "step": 5157
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.660188687849009e-08,
+      "loss": 0.5216,
+      "step": 5158
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.608118376635883e-08,
+      "loss": 0.5171,
+      "step": 5159
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.556420433701191e-08,
+      "loss": 0.4956,
+      "step": 5160
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.505094878365145e-08,
+      "loss": 0.5128,
+      "step": 5161
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4541417298091794e-08,
+      "loss": 0.5355,
+      "step": 5162
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.403561007075173e-08,
+      "loss": 0.526,
+      "step": 5163
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3533527290661175e-08,
+      "loss": 0.5177,
+      "step": 5164
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.303516914545557e-08,
+      "loss": 0.5333,
+      "step": 5165
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.254053582138039e-08,
+      "loss": 0.5334,
+      "step": 5166
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.204962750328777e-08,
+      "loss": 0.4637,
+      "step": 5167
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.156244437463873e-08,
+      "loss": 0.5231,
+      "step": 5168
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.107898661750319e-08,
+      "loss": 0.5297,
+      "step": 5169
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.05992544125544e-08,
+      "loss": 0.4806,
+      "step": 5170
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0123247939078946e-08,
+      "loss": 0.5468,
+      "step": 5171
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9650967374966753e-08,
+      "loss": 0.4718,
+      "step": 5172
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.918241289671664e-08,
+      "loss": 0.5077,
+      "step": 5173
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8717584679437416e-08,
+      "loss": 0.5334,
+      "step": 5174
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.825648289684124e-08,
+      "loss": 0.5092,
+      "step": 5175
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7799107721250272e-08,
+      "loss": 0.4827,
+      "step": 5176
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.734545932359334e-08,
+      "loss": 0.4938,
+      "step": 5177
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.689553787340593e-08,
+      "loss": 0.5154,
+      "step": 5178
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6449343538831328e-08,
+      "loss": 0.501,
+      "step": 5179
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6006876486620593e-08,
+      "loss": 0.4983,
+      "step": 5180
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5568136882129223e-08,
+      "loss": 0.5212,
+      "step": 5181
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5133124889323847e-08,
+      "loss": 0.5134,
+      "step": 5182
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4701840670774414e-08,
+      "loss": 0.5247,
+      "step": 5183
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4274284387658665e-08,
+      "loss": 0.5322,
+      "step": 5184
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3850456199763226e-08,
+      "loss": 0.4952,
+      "step": 5185
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.343035626547696e-08,
+      "loss": 0.5042,
+      "step": 5186
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3013984741800944e-08,
+      "loss": 0.5601,
+      "step": 5187
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.260134178433848e-08,
+      "loss": 0.538,
+      "step": 5188
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2192427547301777e-08,
+      "loss": 0.5297,
+      "step": 5189
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1787242183509694e-08,
+      "loss": 0.5345,
+      "step": 5190
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1385785844385554e-08,
+      "loss": 0.5278,
+      "step": 5191
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0988058679960454e-08,
+      "loss": 0.4957,
+      "step": 5192
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0594060838871055e-08,
+      "loss": 0.5107,
+      "step": 5193
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0203792468362904e-08,
+      "loss": 0.5477,
+      "step": 5194
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.981725371428378e-08,
+      "loss": 0.5391,
+      "step": 5195
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.943444472109146e-08,
+      "loss": 0.4998,
+      "step": 5196
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9055365631847067e-08,
+      "loss": 0.5436,
+      "step": 5197
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8680016588218385e-08,
+      "loss": 0.5262,
+      "step": 5198
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8308397730479875e-08,
+      "loss": 0.5188,
+      "step": 5199
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.794050919751267e-08,
+      "loss": 0.5392,
+      "step": 5200
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7576351126801227e-08,
+      "loss": 0.5117,
+      "step": 5201
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7215923654438916e-08,
+      "loss": 0.5005,
+      "step": 5202
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.685922691512354e-08,
+      "loss": 0.4774,
+      "step": 5203
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6506261042157355e-08,
+      "loss": 0.5187,
+      "step": 5204
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.615702616745041e-08,
+      "loss": 0.5309,
+      "step": 5205
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5811522421517177e-08,
+      "loss": 0.5096,
+      "step": 5206
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.546974993347883e-08,
+      "loss": 0.5138,
+      "step": 5207
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5131708831062074e-08,
+      "loss": 0.5244,
+      "step": 5208
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4797399240595867e-08,
+      "loss": 0.4854,
+      "step": 5209
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4466821287020261e-08,
+      "loss": 0.551,
+      "step": 5210
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4139975093875325e-08,
+      "loss": 0.506,
+      "step": 5211
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3816860783311126e-08,
+      "loss": 0.5225,
+      "step": 5212
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3497478476078852e-08,
+      "loss": 0.4878,
+      "step": 5213
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.318182829153858e-08,
+      "loss": 0.5345,
+      "step": 5214
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2869910347653725e-08,
+      "loss": 0.5386,
+      "step": 5215
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2561724760992156e-08,
+      "loss": 0.4857,
+      "step": 5216
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2257271646728408e-08,
+      "loss": 0.5291,
+      "step": 5217
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1956551118643688e-08,
+      "loss": 0.5322,
+      "step": 5218
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1659563289119213e-08,
+      "loss": 0.54,
+      "step": 5219
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1366308269146198e-08,
+      "loss": 0.5242,
+      "step": 5220
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1076786168318088e-08,
+      "loss": 0.5359,
+      "step": 5221
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0790997094834998e-08,
+      "loss": 0.4988,
+      "step": 5222
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0508941155500386e-08,
+      "loss": 0.528,
+      "step": 5223
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0230618455724372e-08,
+      "loss": 0.5016,
+      "step": 5224
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.956029099518205e-09,
+      "loss": 0.5071,
+      "step": 5225
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.685173189503127e-09,
+      "loss": 0.546,
+      "step": 5226
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.418050826901503e-09,
+      "loss": 0.5249,
+      "step": 5227
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.154662111541257e-09,
+      "loss": 0.5022,
+      "step": 5228
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.895007141855872e-09,
+      "loss": 0.4812,
+      "step": 5229
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.63908601488217e-09,
+      "loss": 0.4702,
+      "step": 5230
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.386898826261424e-09,
+      "loss": 0.5614,
+      "step": 5231
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.13844567024158e-09,
+      "loss": 0.5149,
+      "step": 5232
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.893726639672805e-09,
+      "loss": 0.5413,
+      "step": 5233
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.652741826010835e-09,
+      "loss": 0.5073,
+      "step": 5234
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.415491319316959e-09,
+      "loss": 0.5427,
+      "step": 5235
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.181975208253589e-09,
+      "loss": 0.5386,
+      "step": 5236
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.952193580090915e-09,
+      "loss": 0.5435,
+      "step": 5237
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.72614652070247e-09,
+      "loss": 0.5242,
+      "step": 5238
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.503834114565122e-09,
+      "loss": 0.4729,
+      "step": 5239
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.285256444760191e-09,
+      "loss": 0.4786,
+      "step": 5240
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.070413592974556e-09,
+      "loss": 0.5292,
+      "step": 5241
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.859305639498436e-09,
+      "loss": 0.5334,
+      "step": 5242
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.6519326632265005e-09,
+      "loss": 0.4978,
+      "step": 5243
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.448294741656757e-09,
+      "loss": 0.4843,
+      "step": 5244
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.248391950893883e-09,
+      "loss": 0.5186,
+      "step": 5245
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.052224365642566e-09,
+      "loss": 0.5378,
+      "step": 5246
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8597920592152735e-09,
+      "loss": 0.4943,
+      "step": 5247
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.671095103527812e-09,
+      "loss": 0.5303,
+      "step": 5248
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4861335690982165e-09,
+      "loss": 0.5336,
+      "step": 5249
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3049075250500835e-09,
+      "loss": 0.5249,
+      "step": 5250
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.1274170391103485e-09,
+      "loss": 0.4746,
+      "step": 5251
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.9536621776103955e-09,
+      "loss": 0.5361,
+      "step": 5252
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.783643005484949e-09,
+      "loss": 0.5287,
+      "step": 5253
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6173595862742938e-09,
+      "loss": 0.4945,
+      "step": 5254
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4548119821198324e-09,
+      "loss": 0.4719,
+      "step": 5255
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2960002537685276e-09,
+      "loss": 0.5175,
+      "step": 5256
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1409244605717927e-09,
+      "loss": 0.4952,
+      "step": 5257
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.989584660483269e-09,
+      "loss": 0.4888,
+      "step": 5258
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8419809100610487e-09,
+      "loss": 0.4896,
+      "step": 5259
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6981132644687824e-09,
+      "loss": 0.5376,
+      "step": 5260
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.557981777470131e-09,
+      "loss": 0.5,
+      "step": 5261
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4215865014354245e-09,
+      "loss": 0.5548,
+      "step": 5262
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2889274873372225e-09,
+      "loss": 0.5236,
+      "step": 5263
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1600047847536442e-09,
+      "loss": 0.5234,
+      "step": 5264
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0348184418650386e-09,
+      "loss": 0.4819,
+      "step": 5265
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.913368505455093e-09,
+      "loss": 0.5179,
+      "step": 5266
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7956550209108358e-09,
+      "loss": 0.4436,
+      "step": 5267
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.681678032225964e-09,
+      "loss": 0.536,
+      "step": 5268
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.571437581994184e-09,
+      "loss": 0.4861,
+      "step": 5269
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4649337114136519e-09,
+      "loss": 0.5631,
+      "step": 5270
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3621664602880834e-09,
+      "loss": 0.53,
+      "step": 5271
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2631358670212036e-09,
+      "loss": 0.5455,
+      "step": 5272
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.167841968625627e-09,
+      "loss": 0.5114,
+      "step": 5273
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0762848007106474e-09,
+      "loss": 0.5101,
+      "step": 5274
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.884643974955588e-10,
+      "loss": 0.5326,
+      "step": 5275
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.043807917996639e-10,
+      "loss": 0.4891,
+      "step": 5276
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.240340150444948e-10,
+      "loss": 0.5319,
+      "step": 5277
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.474240972593638e-10,
+      "loss": 0.4889,
+      "step": 5278
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.745510670747025e-10,
+      "loss": 0.4673,
+      "step": 5279
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.054149517220609e-10,
+      "loss": 0.5215,
+      "step": 5280
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.400157770407699e-10,
+      "loss": 0.4757,
+      "step": 5281
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.783535674701689e-10,
+      "loss": 0.5233,
+      "step": 5282
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.204283460562675e-10,
+      "loss": 0.5458,
+      "step": 5283
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.6624013444508387e-10,
+      "loss": 0.5244,
+      "step": 5284
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.157889528881963e-10,
+      "loss": 0.5406,
+      "step": 5285
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6907482024052246e-10,
+      "loss": 0.5664,
+      "step": 5286
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2609775395920907e-10,
+      "loss": 0.5214,
+      "step": 5287
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.868577701058527e-10,
+      "loss": 0.5391,
+      "step": 5288
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5135488334538927e-10,
+      "loss": 0.4914,
+      "step": 5289
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1958910694498394e-10,
+      "loss": 0.6121,
+      "step": 5290
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.15604527773617e-11,
+      "loss": 0.4938,
+      "step": 5291
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.726893131525636e-11,
+      "loss": 0.5291,
+      "step": 5292
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.6714551639182037e-11,
+      "loss": 0.54,
+      "step": 5293
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.98973214285514e-11,
+      "loss": 0.5435,
+      "step": 5294
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.68172469694472e-11,
+      "loss": 0.5459,
+      "step": 5295
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.474333150181424e-12,
+      "loss": 0.5129,
+      "step": 5296
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.868583462405482e-12,
+      "loss": 0.5084,
+      "step": 5297
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.5154,
+      "step": 5298
+    },
+    {
+      "epoch": 3.0,
+      "step": 5298,
+      "total_flos": 5132789872852992.0,
+      "train_loss": 0.7761740333496647,
+      "train_runtime": 16014.7306,
+      "train_samples_per_second": 42.356,
+      "train_steps_per_second": 0.331
+    }
+  ],
+  "max_steps": 5298,
+  "num_train_epochs": 3,
+  "total_flos": 5132789872852992.0,
+  "trial_name": null,
+  "trial_params": null
+}