diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10069 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9995148710676568,
+  "global_step": 1674,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9607843137254903e-05,
+      "loss": 8.2368,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9215686274509805e-05,
+      "loss": 8.2871,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.882352941176471e-05,
+      "loss": 5.2256,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.843137254901961e-05,
+      "loss": 4.6724,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.80392156862745e-05,
+      "loss": 4.0066,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011764705882352942,
+      "loss": 3.7144,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013725490196078434,
+      "loss": 3.6567,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015686274509803922,
+      "loss": 3.4658,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017647058823529413,
+      "loss": 3.1938,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000196078431372549,
+      "loss": 3.0947,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021568627450980395,
+      "loss": 2.7737,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023529411764705883,
+      "loss": 2.3811,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025490196078431374,
+      "loss": 2.1094,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002745098039215687,
+      "loss": 1.8011,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00029411764705882356,
+      "loss": 1.5516,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031372549019607844,
+      "loss": 1.2029,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003333333333333333,
+      "loss": 0.9166,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035294117647058826,
+      "loss": 0.7005,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037254901960784314,
+      "loss": 0.5309,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000392156862745098,
+      "loss": 0.4241,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004117647058823529,
+      "loss": 0.3746,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004313725490196079,
+      "loss": 0.2928,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004509803921568628,
+      "loss": 0.2555,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047058823529411766,
+      "loss": 0.2639,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004901960784313725,
+      "loss": 0.2027,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005098039215686275,
+      "loss": 0.2297,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005294117647058824,
+      "loss": 0.211,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005490196078431374,
+      "loss": 0.1974,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005686274509803922,
+      "loss": 0.1799,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005882352941176471,
+      "loss": 0.168,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006078431372549019,
+      "loss": 0.1746,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006274509803921569,
+      "loss": 0.1649,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006470588235294118,
+      "loss": 0.1437,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006666666666666666,
+      "loss": 0.1436,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006862745098039216,
+      "loss": 0.1388,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007058823529411765,
+      "loss": 0.1432,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007254901960784313,
+      "loss": 0.1367,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007450980392156863,
+      "loss": 0.1281,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007647058823529411,
+      "loss": 0.1407,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000784313725490196,
+      "loss": 0.1402,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000803921568627451,
+      "loss": 0.1272,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008235294117647058,
+      "loss": 0.1299,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008431372549019609,
+      "loss": 0.1308,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008627450980392158,
+      "loss": 0.1288,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008823529411764706,
+      "loss": 0.1199,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009019607843137256,
+      "loss": 0.1213,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009215686274509804,
+      "loss": 0.1199,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009411764705882353,
+      "loss": 0.1214,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009607843137254903,
+      "loss": 0.1147,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000980392156862745,
+      "loss": 0.1158,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 0.1211,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00099999906329556,
+      "loss": 0.1102,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999962531857497,
+      "loss": 0.1166,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999915696810979,
+      "loss": 0.1074,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999985012799153,
+      "loss": 0.114,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999765825644824,
+      "loss": 0.1126,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999662790086726,
+      "loss": 0.1103,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999541021703292,
+      "loss": 0.1138,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999400520950765,
+      "loss": 0.115,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999241288355576,
+      "loss": 0.1332,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999063324514343,
+      "loss": 0.132,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999886663009386,
+      "loss": 0.1389,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998651205831112,
+      "loss": 0.1229,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998417052533245,
+      "loss": 0.1262,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998164171077595,
+      "loss": 0.1164,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999789256241166,
+      "loss": 0.1276,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999760222755311,
+      "loss": 0.1238,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997293167589778,
+      "loss": 0.1212,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999696538367965,
+      "loss": 0.1192,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996618877050878,
+      "loss": 0.1138,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999625364900176,
+      "loss": 0.1146,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995869700900732,
+      "loss": 0.1136,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995467034186383,
+      "loss": 0.1111,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999504565036743,
+      "loss": 0.1057,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994605551022725,
+      "loss": 0.1134,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009994146737801234,
+      "loss": 0.1133,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999366921242205,
+      "loss": 0.1121,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993172976674374,
+      "loss": 0.1092,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999265803241751,
+      "loss": 0.1033,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992124381580857,
+      "loss": 0.1064,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991572026163916,
+      "loss": 0.1067,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991000968236255,
+      "loss": 0.1036,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990411209937523,
+      "loss": 0.1075,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989802753477442,
+      "loss": 0.1096,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989175601135786,
+      "loss": 0.107,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988529755262379,
+      "loss": 0.1064,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987865218277089,
+      "loss": 0.1075,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987181992669812,
+      "loss": 0.1021,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986480081000473,
+      "loss": 0.1032,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985759485899009,
+      "loss": 0.1074,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985020210065353,
+      "loss": 0.0989,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984262256269442,
+      "loss": 0.0988,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983485627351187,
+      "loss": 0.103,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982690326220476,
+      "loss": 0.1033,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998187635585716,
+      "loss": 0.1048,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981043719311034,
+      "loss": 0.0999,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980192419701836,
+      "loss": 0.103,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979322460219233,
+      "loss": 0.0959,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978433844122802,
+      "loss": 0.0987,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977526574742029,
+      "loss": 0.1005,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976600655476281,
+      "loss": 0.0992,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975656089794816,
+      "loss": 0.1026,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974692881236744,
+      "loss": 0.0968,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973711033411032,
+      "loss": 0.0994,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997271054999649,
+      "loss": 0.0979,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971691434741742,
+      "loss": 0.1003,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997065369146523,
+      "loss": 0.1051,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969597324055186,
+      "loss": 0.0974,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996852233646963,
+      "loss": 0.1028,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996742873273634,
+      "loss": 0.0985,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966316516952854,
+      "loss": 0.0982,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996518569328644,
+      "loss": 0.104,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964036265974088,
+      "loss": 0.1028,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962868239322496,
+      "loss": 0.1,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996168161770804,
+      "loss": 0.1042,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960476405576782,
+      "loss": 0.1034,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959252607444427,
+      "loss": 0.0992,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958010227896327,
+      "loss": 0.1024,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956749271587452,
+      "loss": 0.1035,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995546974324237,
+      "loss": 0.1072,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995417164765525,
+      "loss": 0.1071,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952854989689812,
+      "loss": 0.1048,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951519774279334,
+      "loss": 0.1052,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950166006426628,
+      "loss": 0.1007,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948793691204015,
+      "loss": 0.0997,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009947402833753308,
+      "loss": 0.0988,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945993439285797,
+      "loss": 0.0998,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944565513082227,
+      "loss": 0.1004,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943119060492776,
+      "loss": 0.0997,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994165408693704,
+      "loss": 0.1016,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940170597904005,
+      "loss": 0.1013,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993866859895204,
+      "loss": 0.1001,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937148095708855,
+      "loss": 0.101,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099356090938715,
+      "loss": 0.1014,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934051599206337,
+      "loss": 0.0951,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932475617549015,
+      "loss": 0.1012,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930881154804445,
+      "loss": 0.1031,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929268216946792,
+      "loss": 0.105,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992763681001944,
+      "loss": 0.1028,
+      "step": 139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925986940134974,
+      "loss": 0.1011,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924318613475157,
+      "loss": 0.1041,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922631836290898,
+      "loss": 0.1081,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920926614902252,
+      "loss": 0.1014,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919202955698367,
+      "loss": 0.0999,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917460865137484,
+      "loss": 0.1011,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915700349746898,
+      "loss": 0.1005,
+      "step": 146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913921416122937,
+      "loss": 0.1022,
+      "step": 147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912124070930943,
+      "loss": 0.1001,
+      "step": 148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991030832090524,
+      "loss": 0.1037,
+      "step": 149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908474172849113,
+      "loss": 0.1016,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906621633634781,
+      "loss": 0.104,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904750710203372,
+      "loss": 0.1016,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990286140956489,
+      "loss": 0.1045,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900953738798204,
+      "loss": 0.0998,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989902770505101,
+      "loss": 0.0993,
+      "step": 155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897083315539804,
+      "loss": 0.0992,
+      "step": 156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895120577549858,
+      "loss": 0.1,
+      "step": 157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893139498435195,
+      "loss": 0.0976,
+      "step": 158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891140085618556,
+      "loss": 0.1006,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009889122346591378,
+      "loss": 0.104,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988708628891376,
+      "loss": 0.0999,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988503192021444,
+      "loss": 0.0979,
+      "step": 162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882959248190763,
+      "loss": 0.0981,
+      "step": 163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880868280608653,
+      "loss": 0.1018,
+      "step": 164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878759025302587,
+      "loss": 0.0994,
+      "step": 165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876631490175554,
+      "loss": 0.1004,
+      "step": 166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874485683199047,
+      "loss": 0.0957,
+      "step": 167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872321612413011,
+      "loss": 0.102,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870139285925825,
+      "loss": 0.103,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867938711914269,
+      "loss": 0.0996,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986571989862349,
+      "loss": 0.1049,
+      "step": 171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863482854366983,
+      "loss": 0.1032,
+      "step": 172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861227587526539,
+      "loss": 0.1027,
+      "step": 173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858954106552237,
+      "loss": 0.1034,
+      "step": 174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985666241996239,
+      "loss": 0.0997,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009854352536343533,
+      "loss": 0.0993,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852024464350382,
+      "loss": 0.1013,
+      "step": 177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009849678212705795,
+      "loss": 0.098,
+      "step": 178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984731379020075,
+      "loss": 0.1022,
+      "step": 179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844931205694307,
+      "loss": 0.0995,
+      "step": 180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009842530468113578,
+      "loss": 0.1,
+      "step": 181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840111586453687,
+      "loss": 0.1041,
+      "step": 182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009837674569777742,
+      "loss": 0.1008,
+      "step": 183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00098352194272168,
+      "loss": 0.0968,
+      "step": 184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832746167969836,
+      "loss": 0.0986,
+      "step": 185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009830254801303701,
+      "loss": 0.1034,
+      "step": 186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982774533655309,
+      "loss": 0.0984,
+      "step": 187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825217783120512,
+      "loss": 0.1042,
+      "step": 188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822672150476249,
+      "loss": 0.0987,
+      "step": 189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820108448158318,
+      "loss": 0.1058,
+      "step": 190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981752668577245,
+      "loss": 0.0959,
+      "step": 191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981492687299204,
+      "loss": 0.0988,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009812309019558104,
+      "loss": 0.1033,
+      "step": 193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009809673135279268,
+      "loss": 0.0975,
+      "step": 194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980701923003171,
+      "loss": 0.0976,
+      "step": 195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009804347313759126,
+      "loss": 0.1,
+      "step": 196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009801657396472702,
+      "loss": 0.1041,
+      "step": 197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798949488251068,
+      "loss": 0.0974,
+      "step": 198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796223599240262,
+      "loss": 0.0997,
+      "step": 199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979347973965369,
+      "loss": 0.0977,
+      "step": 200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009790717919772101,
+      "loss": 0.1019,
+      "step": 201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009787938149943525,
+      "loss": 0.0985,
+      "step": 202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009785140440583256,
+      "loss": 0.1033,
+      "step": 203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782324802173799,
+      "loss": 0.0969,
+      "step": 204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779491245264838,
+      "loss": 0.0992,
+      "step": 205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009776639780473197,
+      "loss": 0.1003,
+      "step": 206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009773770418482792,
+      "loss": 0.0958,
+      "step": 207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770883170044603,
+      "loss": 0.1043,
+      "step": 208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009767978045976617,
+      "loss": 0.0998,
+      "step": 209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009765055057163812,
+      "loss": 0.0977,
+      "step": 210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762114214558091,
+      "loss": 0.0948,
+      "step": 211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009759155529178255,
+      "loss": 0.102,
+      "step": 212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975617901210996,
+      "loss": 0.1,
+      "step": 213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753184674505672,
+      "loss": 0.1019,
+      "step": 214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750172527584628,
+      "loss": 0.0965,
+      "step": 215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747142582632796,
+      "loss": 0.0975,
+      "step": 216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744094851002825,
+      "loss": 0.1003,
+      "step": 217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974102934411401,
+      "loss": 0.0995,
+      "step": 218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737946073452248,
+      "loss": 0.0995,
+      "step": 219
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734845050569994,
+      "loss": 0.0975,
+      "step": 220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009731726287086211,
+      "loss": 0.0927,
+      "step": 221
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009728589794686341,
+      "loss": 0.1002,
+      "step": 222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009725435585122248,
+      "loss": 0.1029,
+      "step": 223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722263670212182,
+      "loss": 0.1013,
+      "step": 224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009719074061840725,
+      "loss": 0.0996,
+      "step": 225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009715866771958766,
+      "loss": 0.0978,
+      "step": 226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000971264181258343,
+      "loss": 0.0939,
+      "step": 227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009709399195798054,
+      "loss": 0.1027,
+      "step": 228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009706138933752134,
+      "loss": 0.1025,
+      "step": 229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009702861038661273,
+      "loss": 0.1014,
+      "step": 230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969956552280715,
+      "loss": 0.0932,
+      "step": 231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696252398537461,
+      "loss": 0.0981,
+      "step": 232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692921678265882,
+      "loss": 0.0977,
+      "step": 233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968957337447201,
+      "loss": 0.0977,
+      "step": 234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009686207499701334,
+      "loss": 0.0971,
+      "step": 235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682824066565168,
+      "loss": 0.0955,
+      "step": 236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009679423087740624,
+      "loss": 0.0951,
+      "step": 237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676004575970547,
+      "loss": 0.0942,
+      "step": 238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009672568544063481,
+      "loss": 0.0971,
+      "step": 239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009669115004893606,
+      "loss": 0.0971,
+      "step": 240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009665643971400709,
+      "loss": 0.1007,
+      "step": 241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009662155456590116,
+      "loss": 0.1001,
+      "step": 242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009658649473532659,
+      "loss": 0.0979,
+      "step": 243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655126035364618,
+      "loss": 0.0984,
+      "step": 244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009651585155287671,
+      "loss": 0.098,
+      "step": 245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009648026846568853,
+      "loss": 0.0981,
+      "step": 246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009644451122540497,
+      "loss": 0.0985,
+      "step": 247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009640857996600187,
+      "loss": 0.0994,
+      "step": 248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009637247482210716,
+      "loss": 0.096,
+      "step": 249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009633619592900021,
+      "loss": 0.0954,
+      "step": 250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009629974342261141,
+      "loss": 0.0983,
+      "step": 251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009626311743952166,
+      "loss": 0.095,
+      "step": 252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009622631811696188,
+      "loss": 0.0987,
+      "step": 253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009618934559281236,
+      "loss": 0.0946,
+      "step": 254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009615220000560248,
+      "loss": 0.0971,
+      "step": 255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009611488149450995,
+      "loss": 0.0962,
+      "step": 256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009607739019936041,
+      "loss": 0.0998,
+      "step": 257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009603972626062696,
+      "loss": 0.1012,
+      "step": 258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009600188981942947,
+      "loss": 0.1022,
+      "step": 259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009596388101753422,
+      "loss": 0.1014,
+      "step": 260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009592569999735326,
+      "loss": 0.1002,
+      "step": 261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009588734690194389,
+      "loss": 0.1036,
+      "step": 262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000958488218750082,
+      "loss": 0.0997,
+      "step": 263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009581012506089243,
+      "loss": 0.0975,
+      "step": 264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009577125660458649,
+      "loss": 0.1009,
+      "step": 265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009573221665172341,
+      "loss": 0.0978,
+      "step": 266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569300534857876,
+      "loss": 0.0974,
+      "step": 267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009565362284207015,
+      "loss": 0.1006,
+      "step": 268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009561406927975669,
+      "loss": 0.1001,
+      "step": 269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009557434480983833,
+      "loss": 0.1007,
+      "step": 270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009553444958115545,
+      "loss": 0.1005,
+      "step": 271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009549438374318817,
+      "loss": 0.098,
+      "step": 272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009545414744605591,
+      "loss": 0.1009,
+      "step": 273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009541374084051673,
+      "loss": 0.1026,
+      "step": 274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537316407796681,
+      "loss": 0.1013,
+      "step": 275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000953324173104399,
+      "loss": 0.0994,
+      "step": 276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009529150069060669,
+      "loss": 0.0969,
+      "step": 277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009525041437177433,
+      "loss": 0.0962,
+      "step": 278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009520915850788574,
+      "loss": 0.0975,
+      "step": 279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009516773325351915,
+      "loss": 0.1005,
+      "step": 280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009512613876388741,
+      "loss": 0.0981,
+      "step": 281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508437519483752,
+      "loss": 0.1011,
+      "step": 282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009504244270284994,
+      "loss": 0.0984,
+      "step": 283
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000950003414450381,
+      "loss": 0.0962,
+      "step": 284
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009495807157914771,
+      "loss": 0.1079,
+      "step": 285
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009491563326355628,
+      "loss": 0.0983,
+      "step": 286
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009487302665727242,
+      "loss": 0.0998,
+      "step": 287
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009483025191993535,
+      "loss": 0.0989,
+      "step": 288
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009478730921181419,
+      "loss": 0.0984,
+      "step": 289
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009474419869380745,
+      "loss": 0.0984,
+      "step": 290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009470092052744239,
+      "loss": 0.0978,
+      "step": 291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009465747487487439,
+      "loss": 0.0934,
+      "step": 292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009461386189888643,
+      "loss": 0.1006,
+      "step": 293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009457008176288836,
+      "loss": 0.0968,
+      "step": 294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009452613463091637,
+      "loss": 0.0984,
+      "step": 295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009448202066763236,
+      "loss": 0.0975,
+      "step": 296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009443774003832331,
+      "loss": 0.1016,
+      "step": 297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009439329290890068,
+      "loss": 0.1053,
+      "step": 298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009434867944589973,
+      "loss": 0.0978,
+      "step": 299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009430389981647901,
+      "loss": 0.0994,
+      "step": 300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009425895418841961,
+      "loss": 0.0994,
+      "step": 301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421384273012463,
+      "loss": 0.1001,
+      "step": 302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009416856561061846,
+      "loss": 0.1002,
+      "step": 303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009412312299954622,
+      "loss": 0.1016,
+      "step": 304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000940775150671731,
+      "loss": 0.1023,
+      "step": 305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009403174198438371,
+      "loss": 0.0967,
+      "step": 306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009398580392268144,
+      "loss": 0.1001,
+      "step": 307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009393970105418786,
+      "loss": 0.0982,
+      "step": 308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009389343355164199,
+      "loss": 0.0958,
+      "step": 309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009384700158839973,
+      "loss": 0.098,
+      "step": 310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009380040533843319,
+      "loss": 0.0985,
+      "step": 311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009375364497633006,
+      "loss": 0.1004,
+      "step": 312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009370672067729283,
+      "loss": 0.098,
+      "step": 313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009365963261713836,
+      "loss": 0.1013,
+      "step": 314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009361238097229698,
+      "loss": 0.0998,
+      "step": 315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009356496591981204,
+      "loss": 0.0973,
+      "step": 316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009351738763733906,
+      "loss": 0.1009,
+      "step": 317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009346964630314521,
+      "loss": 0.1006,
+      "step": 318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009342174209610857,
+      "loss": 0.1008,
+      "step": 319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009337367519571748,
+      "loss": 0.0951,
+      "step": 320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009332544578206984,
+      "loss": 0.097,
+      "step": 321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009327705403587248,
+      "loss": 0.0977,
+      "step": 322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009322850013844047,
+      "loss": 0.1018,
+      "step": 323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009317978427169638,
+      "loss": 0.1005,
+      "step": 324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009313090661816972,
+      "loss": 0.0936,
+      "step": 325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009308186736099614,
+      "loss": 0.0956,
+      "step": 326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009303266668391679,
+      "loss": 0.0982,
+      "step": 327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009298330477127764,
+      "loss": 0.0986,
+      "step": 328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009293378180802878,
+      "loss": 0.0975,
+      "step": 329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009288409797972376,
+      "loss": 0.0977,
+      "step": 330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009283425347251879,
+      "loss": 0.098,
+      "step": 331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009278424847317217,
+      "loss": 0.0963,
+      "step": 332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009273408316904353,
+      "loss": 0.0982,
+      "step": 333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009268375774809311,
+      "loss": 0.0965,
+      "step": 334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009263327239888109,
+      "loss": 0.0963,
+      "step": 335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009258262731056688,
+      "loss": 0.0994,
+      "step": 336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000925318226729084,
+      "loss": 0.0975,
+      "step": 337
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009248085867626135,
+      "loss": 0.0933,
+      "step": 338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009242973551157857,
+      "loss": 0.1003,
+      "step": 339
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009237845337040919,
+      "loss": 0.0957,
+      "step": 340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000923270124448981,
+      "loss": 0.0961,
+      "step": 341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009227541292778504,
+      "loss": 0.1002,
+      "step": 342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009222365501240402,
+      "loss": 0.0981,
+      "step": 343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009217173889268248,
+      "loss": 0.0992,
+      "step": 344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009211966476314072,
+      "loss": 0.0954,
+      "step": 345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009206743281889097,
+      "loss": 0.0987,
+      "step": 346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000920150432556368,
+      "loss": 0.0966,
+      "step": 347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009196249626967237,
+      "loss": 0.0982,
+      "step": 348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009190979205788169,
+      "loss": 0.0944,
+      "step": 349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009185693081773778,
+      "loss": 0.0972,
+      "step": 350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000918039127473021,
+      "loss": 0.0974,
+      "step": 351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009175073804522371,
+      "loss": 0.0974,
+      "step": 352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009169740691073852,
+      "loss": 0.0942,
+      "step": 353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009164391954366855,
+      "loss": 0.0978,
+      "step": 354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009159027614442126,
+      "loss": 0.0994,
+      "step": 355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009153647691398865,
+      "loss": 0.0955,
+      "step": 356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009148252205394664,
+      "loss": 0.0966,
+      "step": 357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009142841176645428,
+      "loss": 0.0974,
+      "step": 358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009137414625425296,
+      "loss": 0.094,
+      "step": 359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009131972572066563,
+      "loss": 0.0955,
+      "step": 360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009126515036959612,
+      "loss": 0.099,
+      "step": 361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009121042040552836,
+      "loss": 0.0965,
+      "step": 362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009115553603352551,
+      "loss": 0.0935,
+      "step": 363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009110049745922932,
+      "loss": 0.0925,
+      "step": 364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009104530488885931,
+      "loss": 0.0973,
+      "step": 365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009098995852921197,
+      "loss": 0.0972,
+      "step": 366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009093445858766004,
+      "loss": 0.0996,
+      "step": 367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009087880527215166,
+      "loss": 0.0954,
+      "step": 368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009082299879120969,
+      "loss": 0.1019,
+      "step": 369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009076703935393083,
+      "loss": 0.0991,
+      "step": 370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000907109271699849,
+      "loss": 0.0993,
+      "step": 371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009065466244961401,
+      "loss": 0.0993,
+      "step": 372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009059824540363183,
+      "loss": 0.0974,
+      "step": 373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009054167624342274,
+      "loss": 0.098,
+      "step": 374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009048495518094109,
+      "loss": 0.1,
+      "step": 375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009042808242871035,
+      "loss": 0.0944,
+      "step": 376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009037105819982234,
+      "loss": 0.0994,
+      "step": 377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000903138827079365,
+      "loss": 0.1018,
+      "step": 378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009025655616727895,
+      "loss": 0.0938,
+      "step": 379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009019907879264179,
+      "loss": 0.098,
+      "step": 380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009014145079938227,
+      "loss": 0.0974,
+      "step": 381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009008367240342198,
+      "loss": 0.0957,
+      "step": 382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009002574382124605,
+      "loss": 0.0958,
+      "step": 383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008996766526990229,
+      "loss": 0.0962,
+      "step": 384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008990943696700049,
+      "loss": 0.0983,
+      "step": 385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008985105913071148,
+      "loss": 0.0947,
+      "step": 386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008979253197976633,
+      "loss": 0.0978,
+      "step": 387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008973385573345566,
+      "loss": 0.0954,
+      "step": 388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008967503061162865,
+      "loss": 0.097,
+      "step": 389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008961605683469231,
+      "loss": 0.0983,
+      "step": 390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008955693462361064,
+      "loss": 0.0967,
+      "step": 391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008949766419990379,
+      "loss": 0.0954,
+      "step": 392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008943824578564723,
+      "loss": 0.0977,
+      "step": 393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008937867960347094,
+      "loss": 0.0984,
+      "step": 394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008931896587655857,
+      "loss": 0.098,
+      "step": 395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008925910482864652,
+      "loss": 0.1002,
+      "step": 396
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008919909668402325,
+      "loss": 0.098,
+      "step": 397
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008913894166752835,
+      "loss": 0.0977,
+      "step": 398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008907864000455171,
+      "loss": 0.0953,
+      "step": 399
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008901819192103265,
+      "loss": 0.0969,
+      "step": 400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008895759764345915,
+      "loss": 0.1044,
+      "step": 401
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008889685739886691,
+      "loss": 0.0978,
+      "step": 402
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008883597141483854,
+      "loss": 0.0929,
+      "step": 403
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008877493991950276,
+      "loss": 0.0967,
+      "step": 404
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008871376314153343,
+      "loss": 0.0962,
+      "step": 405
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008865244131014882,
+      "loss": 0.0975,
+      "step": 406
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008859097465511064,
+      "loss": 0.0997,
+      "step": 407
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008852936340672324,
+      "loss": 0.1003,
+      "step": 408
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008846760779583273,
+      "loss": 0.0982,
+      "step": 409
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008840570805382618,
+      "loss": 0.0974,
+      "step": 410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008834366441263056,
+      "loss": 0.0963,
+      "step": 411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008828147710471218,
+      "loss": 0.0959,
+      "step": 412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008821914636307546,
+      "loss": 0.0971,
+      "step": 413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000881566724212624,
+      "loss": 0.1017,
+      "step": 414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008809405551335143,
+      "loss": 0.1016,
+      "step": 415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008803129587395673,
+      "loss": 0.1025,
+      "step": 416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000879683937382272,
+      "loss": 0.1002,
+      "step": 417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008790534934184569,
+      "loss": 0.0984,
+      "step": 418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008784216292102807,
+      "loss": 0.0974,
+      "step": 419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008777883471252235,
+      "loss": 0.0953,
+      "step": 420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008771536495360776,
+      "loss": 0.1001,
+      "step": 421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008765175388209394,
+      "loss": 0.0965,
+      "step": 422
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008758800173631997,
+      "loss": 0.0946,
+      "step": 423
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008752410875515353,
+      "loss": 0.0948,
+      "step": 424
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008746007517799,
+      "loss": 0.0956,
+      "step": 425
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008739590124475147,
+      "loss": 0.0952,
+      "step": 426
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008733158719588602,
+      "loss": 0.0976,
+      "step": 427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008726713327236665,
+      "loss": 0.099,
+      "step": 428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008720253971569047,
+      "loss": 0.0998,
+      "step": 429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008713780676787776,
+      "loss": 0.096,
+      "step": 430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008707293467147109,
+      "loss": 0.0978,
+      "step": 431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008700792366953436,
+      "loss": 0.1001,
+      "step": 432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008694277400565198,
+      "loss": 0.0971,
+      "step": 433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008687748592392785,
+      "loss": 0.0972,
+      "step": 434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008681205966898451,
+      "loss": 0.0981,
+      "step": 435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008674649548596221,
+      "loss": 0.0916,
+      "step": 436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008668079362051801,
+      "loss": 0.0977,
+      "step": 437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008661495431882483,
+      "loss": 0.097,
+      "step": 438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000865489778275705,
+      "loss": 0.096,
+      "step": 439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008648286439395696,
+      "loss": 0.0917,
+      "step": 440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008641661426569916,
+      "loss": 0.0988,
+      "step": 441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008635022769102427,
+      "loss": 0.0963,
+      "step": 442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008628370491867068,
+      "loss": 0.0956,
+      "step": 443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008621704619788711,
+      "loss": 0.0977,
+      "step": 444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008615025177843163,
+      "loss": 0.0938,
+      "step": 445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008608332191057075,
+      "loss": 0.099,
+      "step": 446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000860162568450785,
+      "loss": 0.096,
+      "step": 447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008594905683323543,
+      "loss": 0.0983,
+      "step": 448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008588172212682779,
+      "loss": 0.0919,
+      "step": 449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008581425297814641,
+      "loss": 0.0954,
+      "step": 450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000857466496399859,
+      "loss": 0.0931,
+      "step": 451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008567891236564368,
+      "loss": 0.0979,
+      "step": 452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008561104140891894,
+      "loss": 0.1006,
+      "step": 453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008554303702411179,
+      "loss": 0.1012,
+      "step": 454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008547489946602227,
+      "loss": 0.0958,
+      "step": 455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000854066289899494,
+      "loss": 0.0974,
+      "step": 456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008533822585169021,
+      "loss": 0.0961,
+      "step": 457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008526969030753879,
+      "loss": 0.0973,
+      "step": 458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008520102261428533,
+      "loss": 0.0932,
+      "step": 459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008513222302921517,
+      "loss": 0.0962,
+      "step": 460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008506329181010782,
+      "loss": 0.0952,
+      "step": 461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008499422921523596,
+      "loss": 0.0959,
+      "step": 462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008492503550336461,
+      "loss": 0.097,
+      "step": 463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008485571093374996,
+      "loss": 0.0968,
+      "step": 464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008478625576613853,
+      "loss": 0.0979,
+      "step": 465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000847166702607662,
+      "loss": 0.0939,
+      "step": 466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008464695467835717,
+      "loss": 0.0982,
+      "step": 467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008457710928012301,
+      "loss": 0.0962,
+      "step": 468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008450713432776171,
+      "loss": 0.0942,
+      "step": 469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008443703008345668,
+      "loss": 0.0951,
+      "step": 470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008436679680987571,
+      "loss": 0.0959,
+      "step": 471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008429643477017011,
+      "loss": 0.0943,
+      "step": 472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008422594422797361,
+      "loss": 0.0948,
+      "step": 473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000841553254474014,
+      "loss": 0.0921,
+      "step": 474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008408457869304922,
+      "loss": 0.0962,
+      "step": 475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008401370422999224,
+      "loss": 0.0978,
+      "step": 476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008394270232378418,
+      "loss": 0.0946,
+      "step": 477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008387157324045622,
+      "loss": 0.0962,
+      "step": 478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008380031724651608,
+      "loss": 0.0994,
+      "step": 479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00083728934608947,
+      "loss": 0.0938,
+      "step": 480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008365742559520668,
+      "loss": 0.0957,
+      "step": 481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008358579047322639,
+      "loss": 0.0963,
+      "step": 482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008351402951140987,
+      "loss": 0.0965,
+      "step": 483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008344214297863236,
+      "loss": 0.0966,
+      "step": 484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008337013114423961,
+      "loss": 0.0994,
+      "step": 485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008329799427804682,
+      "loss": 0.0956,
+      "step": 486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008322573265033773,
+      "loss": 0.0995,
+      "step": 487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008315334653186343,
+      "loss": 0.0968,
+      "step": 488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008308083619384155,
+      "loss": 0.0972,
+      "step": 489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008300820190795509,
+      "loss": 0.098,
+      "step": 490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008293544394635149,
+      "loss": 0.0962,
+      "step": 491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008286256258164159,
+      "loss": 0.0978,
+      "step": 492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008278955808689857,
+      "loss": 0.0982,
+      "step": 493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008271643073565696,
+      "loss": 0.102,
+      "step": 494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008264318080191162,
+      "loss": 0.0981,
+      "step": 495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008256980856011671,
+      "loss": 0.0961,
+      "step": 496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008249631428518465,
+      "loss": 0.0952,
+      "step": 497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008242269825248508,
+      "loss": 0.0968,
+      "step": 498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008234896073784388,
+      "loss": 0.0969,
+      "step": 499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008227510201754207,
+      "loss": 0.0971,
+      "step": 500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008220112236831481,
+      "loss": 0.099,
+      "step": 501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008212702206735036,
+      "loss": 0.0932,
+      "step": 502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008205280139228906,
+      "loss": 0.095,
+      "step": 503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008197846062122223,
+      "loss": 0.0941,
+      "step": 504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000819040000326912,
+      "loss": 0.0984,
+      "step": 505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008182941990568627,
+      "loss": 0.0923,
+      "step": 506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008175472051964552,
+      "loss": 0.0951,
+      "step": 507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008167990215445396,
+      "loss": 0.0992,
+      "step": 508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008160496509044237,
+      "loss": 0.0975,
+      "step": 509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008152990960838628,
+      "loss": 0.1014,
+      "step": 510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008145473598950489,
+      "loss": 0.0992,
+      "step": 511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008137944451546006,
+      "loss": 0.0955,
+      "step": 512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008130403546835522,
+      "loss": 0.0967,
+      "step": 513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008122850913073432,
+      "loss": 0.0981,
+      "step": 514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008115286578558081,
+      "loss": 0.0974,
+      "step": 515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008107710571631648,
+      "loss": 0.0979,
+      "step": 516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008100122920680052,
+      "loss": 0.0977,
+      "step": 517
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008092523654132839,
+      "loss": 0.0974,
+      "step": 518
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008084912800463076,
+      "loss": 0.0926,
+      "step": 519
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008077290388187243,
+      "loss": 0.0992,
+      "step": 520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008069656445865131,
+      "loss": 0.0976,
+      "step": 521
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008062011002099729,
+      "loss": 0.0927,
+      "step": 522
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008054354085537125,
+      "loss": 0.0946,
+      "step": 523
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008046685724866387,
+      "loss": 0.0991,
+      "step": 524
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008039005948819466,
+      "loss": 0.0939,
+      "step": 525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008031314786171083,
+      "loss": 0.0955,
+      "step": 526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008023612265738623,
+      "loss": 0.096,
+      "step": 527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008015898416382026,
+      "loss": 0.0972,
+      "step": 528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000800817326700368,
+      "loss": 0.0973,
+      "step": 529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008000436846548314,
+      "loss": 0.0958,
+      "step": 530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007992689184002879,
+      "loss": 0.097,
+      "step": 531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007984930308396463,
+      "loss": 0.0973,
+      "step": 532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007977160248800152,
+      "loss": 0.0945,
+      "step": 533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007969379034326949,
+      "loss": 0.0934,
+      "step": 534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007961586694131643,
+      "loss": 0.0965,
+      "step": 535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007953783257410713,
+      "loss": 0.0961,
+      "step": 536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007945968753402215,
+      "loss": 0.0958,
+      "step": 537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007938143211385671,
+      "loss": 0.097,
+      "step": 538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007930306660681961,
+      "loss": 0.0968,
+      "step": 539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007922459130653213,
+      "loss": 0.0955,
+      "step": 540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007914600650702691,
+      "loss": 0.0945,
+      "step": 541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007906731250274687,
+      "loss": 0.0975,
+      "step": 542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007898850958854411,
+      "loss": 0.094,
+      "step": 543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007890959805967879,
+      "loss": 0.0955,
+      "step": 544
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007883057821181802,
+      "loss": 0.093,
+      "step": 545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007875145034103478,
+      "loss": 0.0944,
+      "step": 546
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007867221474380677,
+      "loss": 0.0962,
+      "step": 547
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007859287171701533,
+      "loss": 0.0966,
+      "step": 548
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007851342155794435,
+      "loss": 0.0973,
+      "step": 549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007843386456427905,
+      "loss": 0.0925,
+      "step": 550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007835420103410504,
+      "loss": 0.0936,
+      "step": 551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007827443126590701,
+      "loss": 0.0925,
+      "step": 552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007819455555856775,
+      "loss": 0.0912,
+      "step": 553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007811457421136701,
+      "loss": 0.0962,
+      "step": 554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007803448752398029,
+      "loss": 0.0912,
+      "step": 555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007795429579647782,
+      "loss": 0.0976,
+      "step": 556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007787399932932338,
+      "loss": 0.0966,
+      "step": 557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007779359842337322,
+      "loss": 0.101,
+      "step": 558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007771309337987486,
+      "loss": 0.0944,
+      "step": 559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007763248450046605,
+      "loss": 0.0971,
+      "step": 560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007755177208717357,
+      "loss": 0.0954,
+      "step": 561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007747095644241209,
+      "loss": 0.0974,
+      "step": 562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007739003786898313,
+      "loss": 0.0967,
+      "step": 563
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007730901667007384,
+      "loss": 0.0984,
+      "step": 564
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007722789314925589,
+      "loss": 0.0981,
+      "step": 565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000771466676104843,
+      "loss": 0.0899,
+      "step": 566
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007706534035809639,
+      "loss": 0.093,
+      "step": 567
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007698391169681055,
+      "loss": 0.0948,
+      "step": 568
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007690238193172512,
+      "loss": 0.0939,
+      "step": 569
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000768207513683173,
+      "loss": 0.0936,
+      "step": 570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007673902031244188,
+      "loss": 0.098,
+      "step": 571
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007665718907033031,
+      "loss": 0.097,
+      "step": 572
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007657525794858926,
+      "loss": 0.095,
+      "step": 573
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007649322725419977,
+      "loss": 0.0936,
+      "step": 574
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007641109729451587,
+      "loss": 0.0945,
+      "step": 575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007632886837726358,
+      "loss": 0.0951,
+      "step": 576
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007624654081053965,
+      "loss": 0.0946,
+      "step": 577
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007616411490281048,
+      "loss": 0.0947,
+      "step": 578
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000760815909629109,
+      "loss": 0.0908,
+      "step": 579
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007599896930004309,
+      "loss": 0.093,
+      "step": 580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007591625022377537,
+      "loss": 0.0938,
+      "step": 581
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007583343404404104,
+      "loss": 0.0946,
+      "step": 582
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007575052107113721,
+      "loss": 0.0918,
+      "step": 583
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007566751161572372,
+      "loss": 0.0965,
+      "step": 584
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007558440598882185,
+      "loss": 0.0974,
+      "step": 585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007550120450181323,
+      "loss": 0.0938,
+      "step": 586
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000754179074664387,
+      "loss": 0.0955,
+      "step": 587
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007533451519479704,
+      "loss": 0.0968,
+      "step": 588
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007525102799934391,
+      "loss": 0.0937,
+      "step": 589
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000751674461928906,
+      "loss": 0.0931,
+      "step": 590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007508377008860293,
+      "loss": 0.0988,
+      "step": 591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00075,
+      "loss": 0.0956,
+      "step": 592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007491613624095307,
+      "loss": 0.0944,
+      "step": 593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007483217912568436,
+      "loss": 0.1001,
+      "step": 594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007474812896876589,
+      "loss": 0.0952,
+      "step": 595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007466398608511827,
+      "loss": 0.0958,
+      "step": 596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007457975079000953,
+      "loss": 0.0967,
+      "step": 597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007449542339905402,
+      "loss": 0.0951,
+      "step": 598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007441100422821106,
+      "loss": 0.0955,
+      "step": 599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007432649359378393,
+      "loss": 0.0967,
+      "step": 600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007424189181241856,
+      "loss": 0.0965,
+      "step": 601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007415719920110241,
+      "loss": 0.0948,
+      "step": 602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007407241607716327,
+      "loss": 0.0991,
+      "step": 603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007398754275826801,
+      "loss": 0.0972,
+      "step": 604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007390257956242154,
+      "loss": 0.0908,
+      "step": 605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007381752680796547,
+      "loss": 0.0972,
+      "step": 606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007373238481357695,
+      "loss": 0.0963,
+      "step": 607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007364715389826753,
+      "loss": 0.0977,
+      "step": 608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007356183438138188,
+      "loss": 0.0961,
+      "step": 609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007347642658259674,
+      "loss": 0.0964,
+      "step": 610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007339093082191953,
+      "loss": 0.0981,
+      "step": 611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007330534741968729,
+      "loss": 0.0974,
+      "step": 612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007321967669656544,
+      "loss": 0.0958,
+      "step": 613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007313391897354654,
+      "loss": 0.098,
+      "step": 614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007304807457194918,
+      "loss": 0.0953,
+      "step": 615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007296214381341669,
+      "loss": 0.0961,
+      "step": 616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007287612701991595,
+      "loss": 0.0964,
+      "step": 617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000727900245137362,
+      "loss": 0.0946,
+      "step": 618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007270383661748787,
+      "loss": 0.0931,
+      "step": 619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007261756365410126,
+      "loss": 0.094,
+      "step": 620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007253120594682546,
+      "loss": 0.0979,
+      "step": 621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007244476381922708,
+      "loss": 0.0949,
+      "step": 622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007235823759518899,
+      "loss": 0.098,
+      "step": 623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007227162759890918,
+      "loss": 0.0969,
+      "step": 624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007218493415489956,
+      "loss": 0.094,
+      "step": 625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007209815758798464,
+      "loss": 0.0927,
+      "step": 626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000720112982233004,
+      "loss": 0.0941,
+      "step": 627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007192435638629307,
+      "loss": 0.0951,
+      "step": 628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007183733240271784,
+      "loss": 0.0947,
+      "step": 629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007175022659863773,
+      "loss": 0.098,
+      "step": 630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007166303930042232,
+      "loss": 0.099,
+      "step": 631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007157577083474652,
+      "loss": 0.0971,
+      "step": 632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007148842152858938,
+      "loss": 0.0979,
+      "step": 633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007140099170923281,
+      "loss": 0.0935,
+      "step": 634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007131348170426042,
+      "loss": 0.0983,
+      "step": 635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007122589184155626,
+      "loss": 0.0954,
+      "step": 636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007113822244930357,
+      "loss": 0.0962,
+      "step": 637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007105047385598358,
+      "loss": 0.0949,
+      "step": 638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007096264639037431,
+      "loss": 0.0955,
+      "step": 639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007087474038154925,
+      "loss": 0.0937,
+      "step": 640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007078675615887618,
+      "loss": 0.0938,
+      "step": 641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007069869405201595,
+      "loss": 0.0969,
+      "step": 642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007061055439092125,
+      "loss": 0.0964,
+      "step": 643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007052233750583531,
+      "loss": 0.0931,
+      "step": 644
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007043404372729073,
+      "loss": 0.0918,
+      "step": 645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007034567338610819,
+      "loss": 0.0954,
+      "step": 646
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007025722681339528,
+      "loss": 0.0963,
+      "step": 647
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007016870434054516,
+      "loss": 0.0936,
+      "step": 648
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007008010629923543,
+      "loss": 0.0964,
+      "step": 649
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006999143302142681,
+      "loss": 0.0974,
+      "step": 650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006990268483936189,
+      "loss": 0.0938,
+      "step": 651
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006981386208556393,
+      "loss": 0.0974,
+      "step": 652
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006972496509283561,
+      "loss": 0.0941,
+      "step": 653
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006963599419425777,
+      "loss": 0.094,
+      "step": 654
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006954694972318815,
+      "loss": 0.0939,
+      "step": 655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006945783201326015,
+      "loss": 0.0982,
+      "step": 656
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006936864139838159,
+      "loss": 0.0975,
+      "step": 657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006927937821273345,
+      "loss": 0.0945,
+      "step": 658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006919004279076861,
+      "loss": 0.0964,
+      "step": 659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006910063546721063,
+      "loss": 0.0947,
+      "step": 660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006901115657705247,
+      "loss": 0.0929,
+      "step": 661
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006892160645555521,
+      "loss": 0.097,
+      "step": 662
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006883198543824682,
+      "loss": 0.0926,
+      "step": 663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006874229386092093,
+      "loss": 0.0962,
+      "step": 664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006865253205963555,
+      "loss": 0.096,
+      "step": 665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006856270037071175,
+      "loss": 0.0929,
+      "step": 666
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006847279913073255,
+      "loss": 0.0931,
+      "step": 667
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006838282867654148,
+      "loss": 0.096,
+      "step": 668
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006829278934524145,
+      "loss": 0.0962,
+      "step": 669
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006820268147419344,
+      "loss": 0.0945,
+      "step": 670
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006811250540101518,
+      "loss": 0.0956,
+      "step": 671
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006802226146358001,
+      "loss": 0.0957,
+      "step": 672
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006793195000001551,
+      "loss": 0.1003,
+      "step": 673
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006784157134870227,
+      "loss": 0.0986,
+      "step": 674
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006775112584827265,
+      "loss": 0.0971,
+      "step": 675
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006766061383760943,
+      "loss": 0.0969,
+      "step": 676
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006757003565584463,
+      "loss": 0.0932,
+      "step": 677
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006747939164235819,
+      "loss": 0.0908,
+      "step": 678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006738868213677672,
+      "loss": 0.093,
+      "step": 679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006729790747897218,
+      "loss": 0.0969,
+      "step": 680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000672070680090607,
+      "loss": 0.0936,
+      "step": 681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006711616406740121,
+      "loss": 0.0966,
+      "step": 682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000670251959945942,
+      "loss": 0.0965,
+      "step": 683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006693416413148046,
+      "loss": 0.0939,
+      "step": 684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006684306881913982,
+      "loss": 0.0985,
+      "step": 685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006675191039888977,
+      "loss": 0.0948,
+      "step": 686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006666068921228433,
+      "loss": 0.0941,
+      "step": 687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006656940560111266,
+      "loss": 0.0981,
+      "step": 688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006647805990739782,
+      "loss": 0.0971,
+      "step": 689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006638665247339546,
+      "loss": 0.0984,
+      "step": 690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006629518364159258,
+      "loss": 0.0984,
+      "step": 691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006620365375470623,
+      "loss": 0.0947,
+      "step": 692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006611206315568222,
+      "loss": 0.093,
+      "step": 693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006602041218769382,
+      "loss": 0.0973,
+      "step": 694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006592870119414052,
+      "loss": 0.0919,
+      "step": 695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006583693051864668,
+      "loss": 0.0947,
+      "step": 696
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000657451005050603,
+      "loss": 0.0934,
+      "step": 697
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006565321149745174,
+      "loss": 0.0961,
+      "step": 698
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006556126384011233,
+      "loss": 0.0926,
+      "step": 699
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000654692578775532,
+      "loss": 0.0938,
+      "step": 700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006537719395450391,
+      "loss": 0.0914,
+      "step": 701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006528507241591121,
+      "loss": 0.1007,
+      "step": 702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006519289360693774,
+      "loss": 0.0916,
+      "step": 703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006510065787296064,
+      "loss": 0.0982,
+      "step": 704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006500836555957045,
+      "loss": 0.0938,
+      "step": 705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006491601701256965,
+      "loss": 0.0947,
+      "step": 706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006482361257797139,
+      "loss": 0.096,
+      "step": 707
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006473115260199824,
+      "loss": 0.0969,
+      "step": 708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006463863743108089,
+      "loss": 0.0935,
+      "step": 709
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006454606741185685,
+      "loss": 0.0962,
+      "step": 710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006445344289116909,
+      "loss": 0.0923,
+      "step": 711
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000643607642160648,
+      "loss": 0.0965,
+      "step": 712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000642680317337941,
+      "loss": 0.0947,
+      "step": 713
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006417524579180872,
+      "loss": 0.094,
+      "step": 714
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006408240673776064,
+      "loss": 0.0978,
+      "step": 715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006398951491950089,
+      "loss": 0.098,
+      "step": 716
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006389657068507818,
+      "loss": 0.0938,
+      "step": 717
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006380357438273763,
+      "loss": 0.096,
+      "step": 718
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006371052636091942,
+      "loss": 0.0944,
+      "step": 719
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006361742696825755,
+      "loss": 0.0956,
+      "step": 720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006352427655357848,
+      "loss": 0.0968,
+      "step": 721
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006343107546589981,
+      "loss": 0.0936,
+      "step": 722
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006333782405442904,
+      "loss": 0.092,
+      "step": 723
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006324452266856225,
+      "loss": 0.0957,
+      "step": 724
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006315117165788268,
+      "loss": 0.0974,
+      "step": 725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000630577713721596,
+      "loss": 0.0929,
+      "step": 726
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006296432216134682,
+      "loss": 0.0957,
+      "step": 727
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006287082437558151,
+      "loss": 0.0928,
+      "step": 728
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006277727836518285,
+      "loss": 0.0961,
+      "step": 729
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000626836844806507,
+      "loss": 0.1002,
+      "step": 730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006259004307266426,
+      "loss": 0.0972,
+      "step": 731
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006249635449208084,
+      "loss": 0.0975,
+      "step": 732
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006240261908993447,
+      "loss": 0.0945,
+      "step": 733
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006230883721743462,
+      "loss": 0.1004,
+      "step": 734
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006221500922596488,
+      "loss": 0.0923,
+      "step": 735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006212113546708166,
+      "loss": 0.0952,
+      "step": 736
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006202721629251278,
+      "loss": 0.0963,
+      "step": 737
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006193325205415629,
+      "loss": 0.0913,
+      "step": 738
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006183924310407905,
+      "loss": 0.0958,
+      "step": 739
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000617451897945155,
+      "loss": 0.0958,
+      "step": 740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006165109247786623,
+      "loss": 0.0945,
+      "step": 741
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006155695150669674,
+      "loss": 0.0951,
+      "step": 742
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000614627672337361,
+      "loss": 0.0918,
+      "step": 743
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006136854001187559,
+      "loss": 0.0939,
+      "step": 744
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006127427019416748,
+      "loss": 0.0984,
+      "step": 745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006117995813382358,
+      "loss": 0.0937,
+      "step": 746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006108560418421397,
+      "loss": 0.0961,
+      "step": 747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006099120869886573,
+      "loss": 0.0959,
+      "step": 748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006089677203146151,
+      "loss": 0.096,
+      "step": 749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006080229453583834,
+      "loss": 0.0959,
+      "step": 750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006070777656598615,
+      "loss": 0.0951,
+      "step": 751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006061321847604655,
+      "loss": 0.0905,
+      "step": 752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006051862062031147,
+      "loss": 0.0927,
+      "step": 753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006042398335322183,
+      "loss": 0.0941,
+      "step": 754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006032930702936626,
+      "loss": 0.0917,
+      "step": 755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006023459200347964,
+      "loss": 0.0958,
+      "step": 756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006013983863044195,
+      "loss": 0.0934,
+      "step": 757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006004504726527678,
+      "loss": 0.0971,
+      "step": 758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005995021826315011,
+      "loss": 0.0959,
+      "step": 759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005985535197936896,
+      "loss": 0.0991,
+      "step": 760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005976044876937997,
+      "loss": 0.0957,
+      "step": 761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005966550898876818,
+      "loss": 0.0951,
+      "step": 762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005957053299325565,
+      "loss": 0.0968,
+      "step": 763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005947552113870013,
+      "loss": 0.0954,
+      "step": 764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005938047378109372,
+      "loss": 0.0912,
+      "step": 765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005928539127656157,
+      "loss": 0.0948,
+      "step": 766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005919027398136045,
+      "loss": 0.0968,
+      "step": 767
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005909512225187759,
+      "loss": 0.0968,
+      "step": 768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005899993644462912,
+      "loss": 0.0998,
+      "step": 769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005890471691625894,
+      "loss": 0.095,
+      "step": 770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005880946402353726,
+      "loss": 0.0944,
+      "step": 771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005871417812335934,
+      "loss": 0.0971,
+      "step": 772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005861885957274407,
+      "loss": 0.0962,
+      "step": 773
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005852350872883267,
+      "loss": 0.0915,
+      "step": 774
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005842812594888737,
+      "loss": 0.0933,
+      "step": 775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000583327115902901,
+      "loss": 0.0947,
+      "step": 776
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005823726601054105,
+      "loss": 0.0976,
+      "step": 777
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005814178956725742,
+      "loss": 0.0943,
+      "step": 778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005804628261817204,
+      "loss": 0.0945,
+      "step": 779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005795074552113204,
+      "loss": 0.1005,
+      "step": 780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005785517863409751,
+      "loss": 0.0923,
+      "step": 781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005775958231514018,
+      "loss": 0.0927,
+      "step": 782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005766395692244202,
+      "loss": 0.0916,
+      "step": 783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005756830281429395,
+      "loss": 0.0966,
+      "step": 784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005747262034909446,
+      "loss": 0.0909,
+      "step": 785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005737690988534836,
+      "loss": 0.0922,
+      "step": 786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005728117178166528,
+      "loss": 0.0923,
+      "step": 787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005718540639675845,
+      "loss": 0.0913,
+      "step": 788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005708961408944333,
+      "loss": 0.0921,
+      "step": 789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005699379521863622,
+      "loss": 0.0926,
+      "step": 790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005689795014335296,
+      "loss": 0.0966,
+      "step": 791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005680207922270762,
+      "loss": 0.0939,
+      "step": 792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005670618281591102,
+      "loss": 0.0937,
+      "step": 793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005661026128226956,
+      "loss": 0.0909,
+      "step": 794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005651431498118372,
+      "loss": 0.0929,
+      "step": 795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005641834427214681,
+      "loss": 0.091,
+      "step": 796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005632234951474361,
+      "loss": 0.0947,
+      "step": 797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005622633106864895,
+      "loss": 0.0905,
+      "step": 798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005613028929362647,
+      "loss": 0.0971,
+      "step": 799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005603422454952719,
+      "loss": 0.0982,
+      "step": 800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005593813719628819,
+      "loss": 0.0937,
+      "step": 801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005584202759393128,
+      "loss": 0.0979,
+      "step": 802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005574589610256163,
+      "loss": 0.0935,
+      "step": 803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005564974308236643,
+      "loss": 0.0948,
+      "step": 804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005555356889361349,
+      "loss": 0.0949,
+      "step": 805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005545737389664999,
+      "loss": 0.0981,
+      "step": 806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005536115845190105,
+      "loss": 0.0945,
+      "step": 807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005526492291986841,
+      "loss": 0.0955,
+      "step": 808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005516866766112908,
+      "loss": 0.0962,
+      "step": 809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005507239303633396,
+      "loss": 0.0975,
+      "step": 810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005497609940620651,
+      "loss": 0.0964,
+      "step": 811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005487978713154144,
+      "loss": 0.0939,
+      "step": 812
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005478345657320328,
+      "loss": 0.0931,
+      "step": 813
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005468710809212507,
+      "loss": 0.0943,
+      "step": 814
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005459074204930703,
+      "loss": 0.0963,
+      "step": 815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005449435880581512,
+      "loss": 0.0949,
+      "step": 816
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005439795872277985,
+      "loss": 0.0917,
+      "step": 817
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005430154216139471,
+      "loss": 0.0962,
+      "step": 818
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005420510948291502,
+      "loss": 0.0964,
+      "step": 819
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005410866104865643,
+      "loss": 0.0974,
+      "step": 820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005401219721999364,
+      "loss": 0.09,
+      "step": 821
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005391571835835907,
+      "loss": 0.0965,
+      "step": 822
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000538192248252414,
+      "loss": 0.0911,
+      "step": 823
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005372271698218433,
+      "loss": 0.0932,
+      "step": 824
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005362619519078514,
+      "loss": 0.0933,
+      "step": 825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005352965981269343,
+      "loss": 0.0923,
+      "step": 826
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005343311120960962,
+      "loss": 0.092,
+      "step": 827
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005333654974328378,
+      "loss": 0.0928,
+      "step": 828
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005323997577551409,
+      "loss": 0.0899,
+      "step": 829
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005314338966814564,
+      "loss": 0.0951,
+      "step": 830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005304679178306893,
+      "loss": 0.0958,
+      "step": 831
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005295018248221868,
+      "loss": 0.0929,
+      "step": 832
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000528535621275723,
+      "loss": 0.0938,
+      "step": 833
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005275693108114868,
+      "loss": 0.0976,
+      "step": 834
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000526602897050067,
+      "loss": 0.0963,
+      "step": 835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00052563638361244,
+      "loss": 0.0924,
+      "step": 836
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005246697741199556,
+      "loss": 0.098,
+      "step": 837
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005237030721943236,
+      "loss": 0.0969,
+      "step": 838
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005227362814575995,
+      "loss": 0.0939,
+      "step": 839
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005217694055321725,
+      "loss": 0.0927,
+      "step": 840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005208024480407499,
+      "loss": 0.0941,
+      "step": 841
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005198354126063459,
+      "loss": 0.0948,
+      "step": 842
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005188683028522654,
+      "loss": 0.0908,
+      "step": 843
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005179011224020928,
+      "loss": 0.0964,
+      "step": 844
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005169338748796766,
+      "loss": 0.0888,
+      "step": 845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005159665639091175,
+      "loss": 0.094,
+      "step": 846
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005149991931147531,
+      "loss": 0.0929,
+      "step": 847
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005140317661211457,
+      "loss": 0.0967,
+      "step": 848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005130642865530675,
+      "loss": 0.0973,
+      "step": 849
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005120967580354887,
+      "loss": 0.0925,
+      "step": 850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005111291841935619,
+      "loss": 0.0949,
+      "step": 851
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005101615686526102,
+      "loss": 0.0937,
+      "step": 852
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005091939150381127,
+      "loss": 0.1028,
+      "step": 853
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005082262269756909,
+      "loss": 0.1932,
+      "step": 854
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005072585080910958,
+      "loss": 0.4543,
+      "step": 855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005062907620101936,
+      "loss": 0.158,
+      "step": 856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005053229923589526,
+      "loss": 0.1292,
+      "step": 857
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005043552027634293,
+      "loss": 0.1594,
+      "step": 858
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005033873968497549,
+      "loss": 0.1466,
+      "step": 859
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005024195782441219,
+      "loss": 0.1259,
+      "step": 860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005014517505727701,
+      "loss": 0.1287,
+      "step": 861
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005004839174619736,
+      "loss": 0.1232,
+      "step": 862
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004995160825380265,
+      "loss": 0.121,
+      "step": 863
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00049854824942723,
+      "loss": 0.1112,
+      "step": 864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004975804217558782,
+      "loss": 0.1143,
+      "step": 865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004966126031502451,
+      "loss": 0.1158,
+      "step": 866
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004956447972365708,
+      "loss": 0.121,
+      "step": 867
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004946770076410475,
+      "loss": 0.1025,
+      "step": 868
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004937092379898065,
+      "loss": 0.1062,
+      "step": 869
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004927414919089044,
+      "loss": 0.1054,
+      "step": 870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004917737730243092,
+      "loss": 0.1048,
+      "step": 871
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004908060849618875,
+      "loss": 0.1034,
+      "step": 872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004898384313473899,
+      "loss": 0.1025,
+      "step": 873
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004888708158064381,
+      "loss": 0.1007,
+      "step": 874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048790324196451134,
+      "loss": 0.0971,
+      "step": 875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048693571344693247,
+      "loss": 0.1022,
+      "step": 876
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048596823387885436,
+      "loss": 0.0963,
+      "step": 877
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048500080688524696,
+      "loss": 0.0993,
+      "step": 878
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004840334360908825,
+      "loss": 0.1017,
+      "step": 879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048306612512032343,
+      "loss": 0.0989,
+      "step": 880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004820988775979074,
+      "loss": 0.0998,
+      "step": 881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004811316971477346,
+      "loss": 0.0971,
+      "step": 882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004801645873936543,
+      "loss": 0.0949,
+      "step": 883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004791975519592501,
+      "loss": 0.103,
+      "step": 884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047823059446782767,
+      "loss": 0.0964,
+      "step": 885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004772637185424005,
+      "loss": 0.0969,
+      "step": 886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047629692780567644,
+      "loss": 0.1022,
+      "step": 887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004753302258800444,
+      "loss": 0.1009,
+      "step": 888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047436361638756,
+      "loss": 0.0998,
+      "step": 889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047339710294993327,
+      "loss": 0.0992,
+      "step": 890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004724306891885134,
+      "loss": 0.0956,
+      "step": 891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004714643787242769,
+      "loss": 0.0945,
+      "step": 892
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047049817517781325,
+      "loss": 0.0988,
+      "step": 893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004695320821693106,
+      "loss": 0.0994,
+      "step": 894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046856610331854373,
+      "loss": 0.0973,
+      "step": 895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046760024224485914,
+      "loss": 0.0988,
+      "step": 896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004666345025671622,
+      "loss": 0.1012,
+      "step": 897
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046566888790390386,
+      "loss": 0.0971,
+      "step": 898
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004647034018730658,
+      "loss": 0.1009,
+      "step": 899
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046373804809214875,
+      "loss": 0.0959,
+      "step": 900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046277283017815687,
+      "loss": 0.0918,
+      "step": 901
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000461807751747586,
+      "loss": 0.0959,
+      "step": 902
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046084281641640946,
+      "loss": 0.0979,
+      "step": 903
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004598780278000637,
+      "loss": 0.0955,
+      "step": 904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045891338951343586,
+      "loss": 0.0946,
+      "step": 905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004579489051708499,
+      "loss": 0.096,
+      "step": 906
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004569845783860528,
+      "loss": 0.0977,
+      "step": 907
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045602041277220163,
+      "loss": 0.1008,
+      "step": 908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045505641194184865,
+      "loss": 0.0976,
+      "step": 909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045409257950692987,
+      "loss": 0.092,
+      "step": 910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004531289190787493,
+      "loss": 0.0951,
+      "step": 911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004521654342679672,
+      "loss": 0.0965,
+      "step": 912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045120212868458566,
+      "loss": 0.0961,
+      "step": 913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004502390059379349,
+      "loss": 0.0946,
+      "step": 914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044927606963666056,
+      "loss": 0.0944,
+      "step": 915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044831332338870935,
+      "loss": 0.1004,
+      "step": 916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004473507708013158,
+      "loss": 0.0948,
+      "step": 917
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044638841548098955,
+      "loss": 0.0955,
+      "step": 918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044542626103350017,
+      "loss": 0.0972,
+      "step": 919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004444643110638653,
+      "loss": 0.0965,
+      "step": 920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044350256917633586,
+      "loss": 0.0996,
+      "step": 921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044254103897438385,
+      "loss": 0.0973,
+      "step": 922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044157972406068727,
+      "loss": 0.0972,
+      "step": 923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004406186280371181,
+      "loss": 0.1005,
+      "step": 924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043965775450472826,
+      "loss": 0.0936,
+      "step": 925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043869710706373547,
+      "loss": 0.0965,
+      "step": 926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043773668931351055,
+      "loss": 0.0974,
+      "step": 927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043677650485256404,
+      "loss": 0.0959,
+      "step": 928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004358165572785318,
+      "loss": 0.0944,
+      "step": 929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004348568501881629,
+      "loss": 0.0958,
+      "step": 930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004338973871773044,
+      "loss": 0.0961,
+      "step": 931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004329381718408899,
+      "loss": 0.0908,
+      "step": 932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043197920777292395,
+      "loss": 0.0956,
+      "step": 933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004310204985664703,
+      "loss": 0.0952,
+      "step": 934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000430062047813638,
+      "loss": 0.0961,
+      "step": 935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042910385910556676,
+      "loss": 0.092,
+      "step": 936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004281459360324156,
+      "loss": 0.0956,
+      "step": 937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042718828218334733,
+      "loss": 0.0967,
+      "step": 938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004262309011465164,
+      "loss": 0.0957,
+      "step": 939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042527379650905537,
+      "loss": 0.0986,
+      "step": 940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004243169718570606,
+      "loss": 0.0948,
+      "step": 941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042336043077557993,
+      "loss": 0.0936,
+      "step": 942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042240417684859824,
+      "loss": 0.097,
+      "step": 943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042144821365902477,
+      "loss": 0.0985,
+      "step": 944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042049254478867973,
+      "loss": 0.097,
+      "step": 945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004195371738182796,
+      "loss": 0.0945,
+      "step": 946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041858210432742596,
+      "loss": 0.0944,
+      "step": 947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041762733989458964,
+      "loss": 0.0926,
+      "step": 948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000416672884097099,
+      "loss": 0.0946,
+      "step": 949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004157187405111264,
+      "loss": 0.0966,
+      "step": 950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041476491271167347,
+      "loss": 0.0964,
+      "step": 951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004138114042725596,
+      "loss": 0.0952,
+      "step": 952
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004128582187664066,
+      "loss": 0.0949,
+      "step": 953
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004119053597646273,
+      "loss": 0.0982,
+      "step": 954
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041095283083741077,
+      "loss": 0.0956,
+      "step": 955
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041000063555370894,
+      "loss": 0.1007,
+      "step": 956
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004090487774812244,
+      "loss": 0.0957,
+      "step": 957
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040809726018639555,
+      "loss": 0.0932,
+      "step": 958
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040714608723438437,
+      "loss": 0.0915,
+      "step": 959
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040619526218906276,
+      "loss": 0.0924,
+      "step": 960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004052447886129986,
+      "loss": 0.0939,
+      "step": 961
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040429467006744356,
+      "loss": 0.0922,
+      "step": 962
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004033449101123183,
+      "loss": 0.0963,
+      "step": 963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004023955123062003,
+      "loss": 0.0954,
+      "step": 964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040144648020631055,
+      "loss": 0.096,
+      "step": 965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040049781736849877,
+      "loss": 0.0969,
+      "step": 966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003995495273472323,
+      "loss": 0.0992,
+      "step": 967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003986016136955806,
+      "loss": 0.0924,
+      "step": 968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003976540799652036,
+      "loss": 0.0923,
+      "step": 969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003967069297063376,
+      "loss": 0.0959,
+      "step": 970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003957601664677816,
+      "loss": 0.0937,
+      "step": 971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003948137937968854,
+      "loss": 0.1007,
+      "step": 972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039386781523953457,
+      "loss": 0.0959,
+      "step": 973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039292223434013847,
+      "loss": 0.0975,
+      "step": 974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003919770546416167,
+      "loss": 0.0973,
+      "step": 975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003910322796853848,
+      "loss": 0.1,
+      "step": 976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039008791301134297,
+      "loss": 0.1005,
+      "step": 977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003891439581578604,
+      "loss": 0.1008,
+      "step": 978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038820041866176444,
+      "loss": 0.0948,
+      "step": 979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003872572980583253,
+      "loss": 0.0919,
+      "step": 980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038631459988124403,
+      "loss": 0.1005,
+      "step": 981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003853723276626392,
+      "loss": 0.0921,
+      "step": 982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038443048493303264,
+      "loss": 0.0928,
+      "step": 983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003834890752213379,
+      "loss": 0.0975,
+      "step": 984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038254810205484507,
+      "loss": 0.0966,
+      "step": 985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003816075689592095,
+      "loss": 0.0934,
+      "step": 986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038066747945843737,
+      "loss": 0.0946,
+      "step": 987
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037972783707487237,
+      "loss": 0.0936,
+      "step": 988
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037878864532918366,
+      "loss": 0.0974,
+      "step": 989
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037784990774035124,
+      "loss": 0.0949,
+      "step": 990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003769116278256538,
+      "loss": 0.0934,
+      "step": 991
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003759738091006555,
+      "loss": 0.095,
+      "step": 992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003750364550791917,
+      "loss": 0.0947,
+      "step": 993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037409956927335764,
+      "loss": 0.0958,
+      "step": 994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003731631551934932,
+      "loss": 0.0911,
+      "step": 995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037222721634817146,
+      "loss": 0.0946,
+      "step": 996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000371291756244185,
+      "loss": 0.0954,
+      "step": 997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037035677838653195,
+      "loss": 0.0954,
+      "step": 998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036942228627840425,
+      "loss": 0.0949,
+      "step": 999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003684882834211732,
+      "loss": 0.0943,
+      "step": 1000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036755477331437757,
+      "loss": 0.0916,
+      "step": 1001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003666217594557096,
+      "loss": 0.0959,
+      "step": 1002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000365689245341002,
+      "loss": 0.0949,
+      "step": 1003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003647572344642155,
+      "loss": 0.0951,
+      "step": 1004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036382573031742456,
+      "loss": 0.0926,
+      "step": 1005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003628947363908057,
+      "loss": 0.0934,
+      "step": 1006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036196425617262385,
+      "loss": 0.0941,
+      "step": 1007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003610342931492182,
+      "loss": 0.0967,
+      "step": 1008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036010485080499127,
+      "loss": 0.0947,
+      "step": 1009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003591759326223937,
+      "loss": 0.091,
+      "step": 1010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003582475420819129,
+      "loss": 0.0929,
+      "step": 1011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035731968266205903,
+      "loss": 0.0941,
+      "step": 1012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000356392357839352,
+      "loss": 0.0966,
+      "step": 1013
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035546557108830925,
+      "loss": 0.0952,
+      "step": 1014
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035453932588143156,
+      "loss": 0.0919,
+      "step": 1015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035361362568919096,
+      "loss": 0.0963,
+      "step": 1016
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003526884739800177,
+      "loss": 0.0939,
+      "step": 1017
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003517638742202862,
+      "loss": 0.0989,
+      "step": 1018
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035083982987430364,
+      "loss": 0.0964,
+      "step": 1019
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003499163444042954,
+      "loss": 0.0935,
+      "step": 1020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003489934212703936,
+      "loss": 0.0915,
+      "step": 1021
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034807106393062277,
+      "loss": 0.0941,
+      "step": 1022
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003471492758408879,
+      "loss": 0.0898,
+      "step": 1023
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003462280604549611,
+      "loss": 0.0933,
+      "step": 1024
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003453074212244681,
+      "loss": 0.0954,
+      "step": 1025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034438736159887664,
+      "loss": 0.0946,
+      "step": 1026
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034346788502548267,
+      "loss": 0.0899,
+      "step": 1027
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003425489949493968,
+      "loss": 0.092,
+      "step": 1028
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003416306948135333,
+      "loss": 0.0934,
+      "step": 1029
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034071298805859486,
+      "loss": 0.0924,
+      "step": 1030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033979587812306194,
+      "loss": 0.0937,
+      "step": 1031
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003388793684431779,
+      "loss": 0.0933,
+      "step": 1032
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033796346245293775,
+      "loss": 0.0928,
+      "step": 1033
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003370481635840744,
+      "loss": 0.0944,
+      "step": 1034
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033613347526604556,
+      "loss": 0.0925,
+      "step": 1035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000335219400926022,
+      "loss": 0.0931,
+      "step": 1036
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033430594398887347,
+      "loss": 0.0886,
+      "step": 1037
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033339310787715667,
+      "loss": 0.0949,
+      "step": 1038
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003324808960111024,
+      "loss": 0.0977,
+      "step": 1039
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003315693118086019,
+      "loss": 0.0912,
+      "step": 1040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033065835868519554,
+      "loss": 0.0911,
+      "step": 1041
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003297480400540581,
+      "loss": 0.0948,
+      "step": 1042
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032883835932598804,
+      "loss": 0.0953,
+      "step": 1043
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003279293199093931,
+      "loss": 0.0915,
+      "step": 1044
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032702092521027815,
+      "loss": 0.0958,
+      "step": 1045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000326113178632233,
+      "loss": 0.0942,
+      "step": 1046
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003252060835764181,
+      "loss": 0.0915,
+      "step": 1047
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003242996434415537,
+      "loss": 0.0945,
+      "step": 1048
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032339386162390585,
+      "loss": 0.0971,
+      "step": 1049
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032248874151727353,
+      "loss": 0.0932,
+      "step": 1050
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032158428651297733,
+      "loss": 0.0943,
+      "step": 1051
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032068049999984497,
+      "loss": 0.0973,
+      "step": 1052
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031977738536419995,
+      "loss": 0.0956,
+      "step": 1053
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003188749459898482,
+      "loss": 0.0929,
+      "step": 1054
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003179731852580657,
+      "loss": 0.0926,
+      "step": 1055
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031707210654758554,
+      "loss": 0.0919,
+      "step": 1056
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003161717132345852,
+      "loss": 0.0918,
+      "step": 1057
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031527200869267447,
+      "loss": 0.0963,
+      "step": 1058
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003143729962928825,
+      "loss": 0.0938,
+      "step": 1059
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031347467940364467,
+      "loss": 0.0932,
+      "step": 1060
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003125770613907909,
+      "loss": 0.0957,
+      "step": 1061
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003116801456175319,
+      "loss": 0.0946,
+      "step": 1062
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031078393544444805,
+      "loss": 0.0922,
+      "step": 1063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003098884342294753,
+      "loss": 0.0975,
+      "step": 1064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030899364532789364,
+      "loss": 0.0958,
+      "step": 1065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000308099572092314,
+      "loss": 0.0925,
+      "step": 1066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003072062178726657,
+      "loss": 0.0924,
+      "step": 1067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030631358601618415,
+      "loss": 0.0916,
+      "step": 1068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030542167986739865,
+      "loss": 0.092,
+      "step": 1069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030453050276811855,
+      "loss": 0.0941,
+      "step": 1070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030364005805742246,
+      "loss": 0.0922,
+      "step": 1071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030275034907164395,
+      "loss": 0.0908,
+      "step": 1072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003018613791443609,
+      "loss": 0.0956,
+      "step": 1073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030097315160638126,
+      "loss": 0.0987,
+      "step": 1074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030008566978573204,
+      "loss": 0.0905,
+      "step": 1075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029919893700764566,
+      "loss": 0.0974,
+      "step": 1076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002983129565945484,
+      "loss": 0.0905,
+      "step": 1077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002974277318660472,
+      "loss": 0.0948,
+      "step": 1078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029654326613891816,
+      "loss": 0.0932,
+      "step": 1079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002956595627270928,
+      "loss": 0.0966,
+      "step": 1080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029477662494164705,
+      "loss": 0.0973,
+      "step": 1081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002938944560907876,
+      "loss": 0.0907,
+      "step": 1082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002930130594798405,
+      "loss": 0.0942,
+      "step": 1083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029213243841123837,
+      "loss": 0.0947,
+      "step": 1084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029125259618450767,
+      "loss": 0.0934,
+      "step": 1085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002903735360962569,
+      "loss": 0.0918,
+      "step": 1086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002894952614401642,
+      "loss": 0.0925,
+      "step": 1087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028861777550696444,
+      "loss": 0.09,
+      "step": 1088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028774108158443755,
+      "loss": 0.0993,
+      "step": 1089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028686518295739595,
+      "loss": 0.0923,
+      "step": 1090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000285990082907672,
+      "loss": 0.0946,
+      "step": 1091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028511578471410637,
+      "loss": 0.0959,
+      "step": 1092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002842422916525349,
+      "loss": 0.0941,
+      "step": 1093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028336960699577694,
+      "loss": 0.0927,
+      "step": 1094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028249773401362266,
+      "loss": 0.094,
+      "step": 1095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028162667597282176,
+      "loss": 0.0957,
+      "step": 1096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028075643613706934,
+      "loss": 0.0959,
+      "step": 1097
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002798870177669961,
+      "loss": 0.0955,
+      "step": 1098
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027901842412015357,
+      "loss": 0.0942,
+      "step": 1099
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027815065845100434,
+      "loss": 0.0963,
+      "step": 1100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027728372401090805,
+      "loss": 0.0973,
+      "step": 1101
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002764176240481102,
+      "loss": 0.0945,
+      "step": 1102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002755523618077294,
+      "loss": 0.0925,
+      "step": 1103
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027468794053174544,
+      "loss": 0.0953,
+      "step": 1104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027382436345898756,
+      "loss": 0.0933,
+      "step": 1105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002729616338251215,
+      "loss": 0.0928,
+      "step": 1106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027209975486263804,
+      "loss": 0.0939,
+      "step": 1107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002712387298008406,
+      "loss": 0.0913,
+      "step": 1108
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002703785618658332,
+      "loss": 0.0984,
+      "step": 1109
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002695192542805081,
+      "loss": 0.0938,
+      "step": 1110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002686608102645347,
+      "loss": 0.0935,
+      "step": 1111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026780323303434583,
+      "loss": 0.0978,
+      "step": 1112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002669465258031273,
+      "loss": 0.0946,
+      "step": 1113
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026609069178080486,
+      "loss": 0.0939,
+      "step": 1114
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026523573417403255,
+      "loss": 0.0984,
+      "step": 1115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026438165618618124,
+      "loss": 0.0957,
+      "step": 1116
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026352846101732475,
+      "loss": 0.0946,
+      "step": 1117
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026267615186423065,
+      "loss": 0.0897,
+      "step": 1118
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002618247319203452,
+      "loss": 0.0899,
+      "step": 1119
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026097420437578444,
+      "loss": 0.0924,
+      "step": 1120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026012457241731985,
+      "loss": 0.0953,
+      "step": 1121
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002592758392283675,
+      "loss": 0.0935,
+      "step": 1122
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025842800798897596,
+      "loss": 0.0927,
+      "step": 1123
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025758108187581445,
+      "loss": 0.0932,
+      "step": 1124
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025673506406216076,
+      "loss": 0.0949,
+      "step": 1125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002558899577178894,
+      "loss": 0.092,
+      "step": 1126
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025504576600945993,
+      "loss": 0.0931,
+      "step": 1127
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002542024920999047,
+      "loss": 0.0952,
+      "step": 1128
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002533601391488175,
+      "loss": 0.0947,
+      "step": 1129
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002525187103123411,
+      "loss": 0.0952,
+      "step": 1130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002516782087431565,
+      "loss": 0.0967,
+      "step": 1131
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025083863759046943,
+      "loss": 0.0942,
+      "step": 1132
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002500000000000001,
+      "loss": 0.0937,
+      "step": 1133
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002491622991139708,
+      "loss": 0.0971,
+      "step": 1134
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024832553807109395,
+      "loss": 0.0974,
+      "step": 1135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002474897200065611,
+      "loss": 0.0919,
+      "step": 1136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002466548480520296,
+      "loss": 0.0941,
+      "step": 1137
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024582092533561326,
+      "loss": 0.0957,
+      "step": 1138
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002449879549818676,
+      "loss": 0.0919,
+      "step": 1139
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024415594011178147,
+      "loss": 0.0956,
+      "step": 1140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002433248838427628,
+      "loss": 0.0925,
+      "step": 1141
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002424947892886279,
+      "loss": 0.0932,
+      "step": 1142
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024166565955958974,
+      "loss": 0.094,
+      "step": 1143
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002408374977622464,
+      "loss": 0.092,
+      "step": 1144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024001030699956916,
+      "loss": 0.0934,
+      "step": 1145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023918409037089111,
+      "loss": 0.0925,
+      "step": 1146
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023835885097189536,
+      "loss": 0.0936,
+      "step": 1147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023753459189460358,
+      "loss": 0.0957,
+      "step": 1148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023671131622736424,
+      "loss": 0.0943,
+      "step": 1149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023588902705484128,
+      "loss": 0.0937,
+      "step": 1150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023506772745800236,
+      "loss": 0.0966,
+      "step": 1151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002342474205141073,
+      "loss": 0.0984,
+      "step": 1152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023342810929669712,
+      "loss": 0.0931,
+      "step": 1153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002326097968755812,
+      "loss": 0.0948,
+      "step": 1154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023179248631682726,
+      "loss": 0.0918,
+      "step": 1155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002309761806827489,
+      "loss": 0.092,
+      "step": 1156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023016088303189446,
+      "loss": 0.0932,
+      "step": 1157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002293465964190362,
+      "loss": 0.093,
+      "step": 1158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022853332389515695,
+      "loss": 0.0978,
+      "step": 1159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022772106850744135,
+      "loss": 0.0936,
+      "step": 1160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022690983329926156,
+      "loss": 0.0927,
+      "step": 1161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002260996213101687,
+      "loss": 0.0897,
+      "step": 1162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022529043557587914,
+      "loss": 0.0967,
+      "step": 1163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022448227912826447,
+      "loss": 0.0912,
+      "step": 1164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022367515499533954,
+      "loss": 0.0961,
+      "step": 1165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002228690662012514,
+      "loss": 0.0954,
+      "step": 1166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022206401576626788,
+      "loss": 0.0941,
+      "step": 1167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022126000670676628,
+      "loss": 0.0931,
+      "step": 1168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022045704203522192,
+      "loss": 0.0915,
+      "step": 1169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021965512476019723,
+      "loss": 0.0956,
+      "step": 1170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021885425788633,
+      "loss": 0.0945,
+      "step": 1171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021805444441432231,
+      "loss": 0.0945,
+      "step": 1172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021725568734093,
+      "loss": 0.093,
+      "step": 1173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021645798965894952,
+      "loss": 0.0909,
+      "step": 1174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002156613543572095,
+      "loss": 0.0936,
+      "step": 1175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002148657844205567,
+      "loss": 0.0909,
+      "step": 1176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021407128282984662,
+      "loss": 0.0977,
+      "step": 1177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002132778525619325,
+      "loss": 0.0955,
+      "step": 1178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002124854965896522,
+      "loss": 0.0925,
+      "step": 1179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021169421788181996,
+      "loss": 0.0961,
+      "step": 1180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021090401940321212,
+      "loss": 0.0929,
+      "step": 1181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021011490411455892,
+      "loss": 0.0917,
+      "step": 1182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020932687497253133,
+      "loss": 0.0927,
+      "step": 1183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000208539934929731,
+      "loss": 0.091,
+      "step": 1184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002077540869346788,
+      "loss": 0.0942,
+      "step": 1185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020696933393180394,
+      "loss": 0.092,
+      "step": 1186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020618567886143297,
+      "loss": 0.0929,
+      "step": 1187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020540312465977862,
+      "loss": 0.0936,
+      "step": 1188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002046216742589288,
+      "loss": 0.0952,
+      "step": 1189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020384133058683586,
+      "loss": 0.0932,
+      "step": 1190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020306209656730523,
+      "loss": 0.0918,
+      "step": 1191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020228397511998464,
+      "loss": 0.097,
+      "step": 1192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002015069691603539,
+      "loss": 0.0931,
+      "step": 1193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020073108159971193,
+      "loss": 0.091,
+      "step": 1194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001999563153451689,
+      "loss": 0.0926,
+      "step": 1195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001991826732996319,
+      "loss": 0.1004,
+      "step": 1196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019841015836179733,
+      "loss": 0.0924,
+      "step": 1197
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019763877342613784,
+      "loss": 0.0939,
+      "step": 1198
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019686852138289164,
+      "loss": 0.093,
+      "step": 1199
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019609940511805353,
+      "loss": 0.0953,
+      "step": 1200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019533142751336125,
+      "loss": 0.0963,
+      "step": 1201
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019456459144628763,
+      "loss": 0.0943,
+      "step": 1202
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000193798899790027,
+      "loss": 0.0912,
+      "step": 1203
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019303435541348697,
+      "loss": 0.0925,
+      "step": 1204
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001922709611812758,
+      "loss": 0.0933,
+      "step": 1205
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001915087199536925,
+      "loss": 0.0983,
+      "step": 1206
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001907476345867162,
+      "loss": 0.0921,
+      "step": 1207
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018998770793199487,
+      "loss": 0.092,
+      "step": 1208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018922894283683535,
+      "loss": 0.0921,
+      "step": 1209
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018847134214419203,
+      "loss": 0.0906,
+      "step": 1210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018771490869265683,
+      "loss": 0.0966,
+      "step": 1211
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018695964531644787,
+      "loss": 0.0923,
+      "step": 1212
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018620555484539947,
+      "loss": 0.0944,
+      "step": 1213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018545264010495104,
+      "loss": 0.095,
+      "step": 1214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018470090391613735,
+      "loss": 0.0942,
+      "step": 1215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018395034909557628,
+      "loss": 0.0914,
+      "step": 1216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018320097845546057,
+      "loss": 0.0896,
+      "step": 1217
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018245279480354505,
+      "loss": 0.0897,
+      "step": 1218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018170580094313738,
+      "loss": 0.0958,
+      "step": 1219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018095999967308803,
+      "loss": 0.0946,
+      "step": 1220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018021539378777768,
+      "loss": 0.0954,
+      "step": 1221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017947198607710967,
+      "loss": 0.0952,
+      "step": 1222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001787297793264965,
+      "loss": 0.0929,
+      "step": 1223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017798877631685202,
+      "loss": 0.0948,
+      "step": 1224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017724897982457943,
+      "loss": 0.0943,
+      "step": 1225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017651039262156127,
+      "loss": 0.0942,
+      "step": 1226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017577301747514923,
+      "loss": 0.0931,
+      "step": 1227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017503685714815358,
+      "loss": 0.0911,
+      "step": 1228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017430191439883296,
+      "loss": 0.0973,
+      "step": 1229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001735681919808839,
+      "loss": 0.0934,
+      "step": 1230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017283569264343057,
+      "loss": 0.0971,
+      "step": 1231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001721044191310145,
+      "loss": 0.0946,
+      "step": 1232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001713743741835842,
+      "loss": 0.0917,
+      "step": 1233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000170645560536485,
+      "loss": 0.0974,
+      "step": 1234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001699179809204493,
+      "loss": 0.0953,
+      "step": 1235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016919163806158454,
+      "loss": 0.0915,
+      "step": 1236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016846653468136587,
+      "loss": 0.0906,
+      "step": 1237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016774267349662275,
+      "loss": 0.094,
+      "step": 1238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001670200572195316,
+      "loss": 0.0945,
+      "step": 1239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016629868855760405,
+      "loss": 0.0921,
+      "step": 1240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001655785702136764,
+      "loss": 0.0937,
+      "step": 1241
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016485970488590152,
+      "loss": 0.0928,
+      "step": 1242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016414209526773615,
+      "loss": 0.0939,
+      "step": 1243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016342574404793327,
+      "loss": 0.0941,
+      "step": 1244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016271065391053013,
+      "loss": 0.0929,
+      "step": 1245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016199682753483924,
+      "loss": 0.0923,
+      "step": 1246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001612842675954379,
+      "loss": 0.0964,
+      "step": 1247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016057297676215832,
+      "loss": 0.0878,
+      "step": 1248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015986295770007764,
+      "loss": 0.0939,
+      "step": 1249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001591542130695079,
+      "loss": 0.0896,
+      "step": 1250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001584467455259861,
+      "loss": 0.0938,
+      "step": 1251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001577405577202641,
+      "loss": 0.092,
+      "step": 1252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015703565229829904,
+      "loss": 0.0915,
+      "step": 1253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001563320319012428,
+      "loss": 0.0962,
+      "step": 1254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015562969916543335,
+      "loss": 0.0934,
+      "step": 1255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015492865672238278,
+      "loss": 0.0923,
+      "step": 1256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015422890719876998,
+      "loss": 0.0946,
+      "step": 1257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015353045321642828,
+      "loss": 0.094,
+      "step": 1258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015283329739233808,
+      "loss": 0.0932,
+      "step": 1259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015213744233861465,
+      "loss": 0.0945,
+      "step": 1260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015144289066250044,
+      "loss": 0.0969,
+      "step": 1261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015074964496635408,
+      "loss": 0.0941,
+      "step": 1262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015005770784764033,
+      "loss": 0.0919,
+      "step": 1263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014936708189892213,
+      "loss": 0.0918,
+      "step": 1264
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014867776970784835,
+      "loss": 0.0933,
+      "step": 1265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001479897738571468,
+      "loss": 0.0934,
+      "step": 1266
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014730309692461218,
+      "loss": 0.0919,
+      "step": 1267
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014661774148309797,
+      "loss": 0.0922,
+      "step": 1268
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014593371010050604,
+      "loss": 0.094,
+      "step": 1269
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001452510053397773,
+      "loss": 0.094,
+      "step": 1270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014456962975888216,
+      "loss": 0.0896,
+      "step": 1271
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001438895859108107,
+      "loss": 0.0914,
+      "step": 1272
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001432108763435633,
+      "loss": 0.0888,
+      "step": 1273
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014253350360014094,
+      "loss": 0.0919,
+      "step": 1274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014185747021853602,
+      "loss": 0.0942,
+      "step": 1275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001411827787317221,
+      "loss": 0.0903,
+      "step": 1276
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014050943166764567,
+      "loss": 0.0896,
+      "step": 1277
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013983743154921503,
+      "loss": 0.0929,
+      "step": 1278
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013916678089429264,
+      "loss": 0.0914,
+      "step": 1279
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001384974822156837,
+      "loss": 0.092,
+      "step": 1280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001378295380211289,
+      "loss": 0.0898,
+      "step": 1281
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013716295081329317,
+      "loss": 0.0891,
+      "step": 1282
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013649772308975733,
+      "loss": 0.0908,
+      "step": 1283
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013583385734300857,
+      "loss": 0.0935,
+      "step": 1284
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013517135606043047,
+      "loss": 0.0916,
+      "step": 1285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013451022172429494,
+      "loss": 0.0914,
+      "step": 1286
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013385045681175178,
+      "loss": 0.0916,
+      "step": 1287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001331920637948199,
+      "loss": 0.0926,
+      "step": 1288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013253504514037796,
+      "loss": 0.0906,
+      "step": 1289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013187940331015502,
+      "loss": 0.0903,
+      "step": 1290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001312251407607216,
+      "loss": 0.0918,
+      "step": 1291
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013057225994348026,
+      "loss": 0.0897,
+      "step": 1292
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001299207633046564,
+      "loss": 0.091,
+      "step": 1293
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012927065328528924,
+      "loss": 0.0938,
+      "step": 1294
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001286219323212225,
+      "loss": 0.0916,
+      "step": 1295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001279746028430953,
+      "loss": 0.0931,
+      "step": 1296
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012732866727633362,
+      "loss": 0.0883,
+      "step": 1297
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012668412804113982,
+      "loss": 0.0937,
+      "step": 1298
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012604098755248538,
+      "loss": 0.0905,
+      "step": 1299
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012539924822010007,
+      "loss": 0.0937,
+      "step": 1300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001247589124484646,
+      "loss": 0.0918,
+      "step": 1301
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001241199826368003,
+      "loss": 0.0971,
+      "step": 1302
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012348246117906063,
+      "loss": 0.0952,
+      "step": 1303
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012284635046392245,
+      "loss": 0.0892,
+      "step": 1304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001222116528747766,
+      "loss": 0.0921,
+      "step": 1305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012157837078971929,
+      "loss": 0.0917,
+      "step": 1306
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001209465065815431,
+      "loss": 0.0891,
+      "step": 1307
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012031606261772804,
+      "loss": 0.0971,
+      "step": 1308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011968704126043278,
+      "loss": 0.0952,
+      "step": 1309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011905944486648568,
+      "loss": 0.0934,
+      "step": 1310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011843327578737611,
+      "loss": 0.0933,
+      "step": 1311
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011780853636924544,
+      "loss": 0.0928,
+      "step": 1312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011718522895287848,
+      "loss": 0.0922,
+      "step": 1313
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011656335587369443,
+      "loss": 0.0913,
+      "step": 1314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011594291946173846,
+      "loss": 0.0909,
+      "step": 1315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011532392204167274,
+      "loss": 0.093,
+      "step": 1316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011470636593276778,
+      "loss": 0.0933,
+      "step": 1317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011409025344889363,
+      "loss": 0.0938,
+      "step": 1318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011347558689851189,
+      "loss": 0.0974,
+      "step": 1319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011286236858466559,
+      "loss": 0.0919,
+      "step": 1320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011225060080497256,
+      "loss": 0.0931,
+      "step": 1321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011164028585161456,
+      "loss": 0.0923,
+      "step": 1322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011103142601133098,
+      "loss": 0.09,
+      "step": 1323
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011042402356540853,
+      "loss": 0.0946,
+      "step": 1324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010981808078967348,
+      "loss": 0.096,
+      "step": 1325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001092135999544831,
+      "loss": 0.0926,
+      "step": 1326
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010861058332471651,
+      "loss": 0.0965,
+      "step": 1327
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010800903315976756,
+      "loss": 0.0895,
+      "step": 1328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010740895171353492,
+      "loss": 0.0917,
+      "step": 1329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010681034123441447,
+      "loss": 0.0903,
+      "step": 1330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010621320396529055,
+      "loss": 0.0915,
+      "step": 1331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010561754214352765,
+      "loss": 0.0946,
+      "step": 1332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010502335800096213,
+      "loss": 0.095,
+      "step": 1333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010443065376389366,
+      "loss": 0.0888,
+      "step": 1334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010383943165307696,
+      "loss": 0.0902,
+      "step": 1335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010324969388371363,
+      "loss": 0.0899,
+      "step": 1336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010266144266544352,
+      "loss": 0.0914,
+      "step": 1337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010207468020233662,
+      "loss": 0.0921,
+      "step": 1338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010148940869288542,
+      "loss": 0.092,
+      "step": 1339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010090563032999505,
+      "loss": 0.092,
+      "step": 1340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010032334730097714,
+      "loss": 0.0933,
+      "step": 1341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.974256178753954e-05,
+      "loss": 0.0902,
+      "step": 1342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.916327596578017e-05,
+      "loss": 0.0975,
+      "step": 1343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.858549200617734e-05,
+      "loss": 0.0913,
+      "step": 1344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.800921207358215e-05,
+      "loss": 0.092,
+      "step": 1345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.743443832721055e-05,
+      "loss": 0.0947,
+      "step": 1346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.686117292063501e-05,
+      "loss": 0.0953,
+      "step": 1347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.628941800177654e-05,
+      "loss": 0.0902,
+      "step": 1348
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.571917571289662e-05,
+      "loss": 0.0896,
+      "step": 1349
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.515044819058921e-05,
+      "loss": 0.092,
+      "step": 1350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.458323756577264e-05,
+      "loss": 0.0913,
+      "step": 1351
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.401754596368178e-05,
+      "loss": 0.093,
+      "step": 1352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.345337550385985e-05,
+      "loss": 0.0934,
+      "step": 1353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.28907283001511e-05,
+      "loss": 0.0948,
+      "step": 1354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.232960646069171e-05,
+      "loss": 0.0932,
+      "step": 1355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.177001208790309e-05,
+      "loss": 0.0925,
+      "step": 1356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.121194727848336e-05,
+      "loss": 0.0961,
+      "step": 1357
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.065541412339956e-05,
+      "loss": 0.0931,
+      "step": 1358
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.010041470788034e-05,
+      "loss": 0.0971,
+      "step": 1359
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.954695111140688e-05,
+      "loss": 0.0898,
+      "step": 1360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.899502540770687e-05,
+      "loss": 0.0964,
+      "step": 1361
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.844463966474491e-05,
+      "loss": 0.0924,
+      "step": 1362
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.789579594471648e-05,
+      "loss": 0.0927,
+      "step": 1363
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.734849630403874e-05,
+      "loss": 0.0936,
+      "step": 1364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.680274279334372e-05,
+      "loss": 0.089,
+      "step": 1365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.625853745747048e-05,
+      "loss": 0.0902,
+      "step": 1366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.571588233545713e-05,
+      "loss": 0.093,
+      "step": 1367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.517477946053353e-05,
+      "loss": 0.0925,
+      "step": 1368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.46352308601136e-05,
+      "loss": 0.091,
+      "step": 1369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.409723855578754e-05,
+      "loss": 0.0936,
+      "step": 1370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.35608045633145e-05,
+      "loss": 0.0916,
+      "step": 1371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.302593089261496e-05,
+      "loss": 0.0921,
+      "step": 1372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.2492619547763e-05,
+      "loss": 0.0916,
+      "step": 1373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.196087252697909e-05,
+      "loss": 0.091,
+      "step": 1374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.143069182262225e-05,
+      "loss": 0.0945,
+      "step": 1375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.090207942118333e-05,
+      "loss": 0.0908,
+      "step": 1376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.037503730327633e-05,
+      "loss": 0.0934,
+      "step": 1377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.984956744363208e-05,
+      "loss": 0.0907,
+      "step": 1378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.932567181109051e-05,
+      "loss": 0.0956,
+      "step": 1379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.880335236859281e-05,
+      "loss": 0.0921,
+      "step": 1380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.82826110731752e-05,
+      "loss": 0.094,
+      "step": 1381
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.776344987595984e-05,
+      "loss": 0.0927,
+      "step": 1382
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.724587072214972e-05,
+      "loss": 0.0915,
+      "step": 1383
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.672987555101907e-05,
+      "loss": 0.0913,
+      "step": 1384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.621546629590814e-05,
+      "loss": 0.0953,
+      "step": 1385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.570264488421447e-05,
+      "loss": 0.0974,
+      "step": 1386
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.519141323738654e-05,
+      "loss": 0.0937,
+      "step": 1387
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.468177327091607e-05,
+      "loss": 0.09,
+      "step": 1388
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.417372689433122e-05,
+      "loss": 0.0946,
+      "step": 1389
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.366727601118911e-05,
+      "loss": 0.0917,
+      "step": 1390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.316242251906896e-05,
+      "loss": 0.0894,
+      "step": 1391
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.26591683095647e-05,
+      "loss": 0.0938,
+      "step": 1392
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.21575152682783e-05,
+      "loss": 0.0914,
+      "step": 1393
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.165746527481215e-05,
+      "loss": 0.0928,
+      "step": 1394
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.115902020276238e-05,
+      "loss": 0.092,
+      "step": 1395
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.066218191971219e-05,
+      "loss": 0.0897,
+      "step": 1396
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.016695228722358e-05,
+      "loss": 0.0902,
+      "step": 1397
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.967333316083224e-05,
+      "loss": 0.0922,
+      "step": 1398
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.918132639003876e-05,
+      "loss": 0.093,
+      "step": 1399
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.869093381830278e-05,
+      "loss": 0.0906,
+      "step": 1400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.820215728303625e-05,
+      "loss": 0.0944,
+      "step": 1401
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.771499861559538e-05,
+      "loss": 0.0913,
+      "step": 1402
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.722945964127525e-05,
+      "loss": 0.091,
+      "step": 1403
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.674554217930162e-05,
+      "loss": 0.0915,
+      "step": 1404
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.626324804282524e-05,
+      "loss": 0.0945,
+      "step": 1405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.578257903891427e-05,
+      "loss": 0.0892,
+      "step": 1406
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.530353696854791e-05,
+      "loss": 0.0935,
+      "step": 1407
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.482612362660944e-05,
+      "loss": 0.0902,
+      "step": 1408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.435034080187968e-05,
+      "loss": 0.0951,
+      "step": 1409
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.387619027703017e-05,
+      "loss": 0.0905,
+      "step": 1410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.340367382861651e-05,
+      "loss": 0.0867,
+      "step": 1411
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.293279322707169e-05,
+      "loss": 0.0953,
+      "step": 1412
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.246355023669959e-05,
+      "loss": 0.0944,
+      "step": 1413
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.19959466156681e-05,
+      "loss": 0.0917,
+      "step": 1414
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.152998411600269e-05,
+      "loss": 0.0923,
+      "step": 1415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.106566448358025e-05,
+      "loss": 0.0943,
+      "step": 1416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.060298945812143e-05,
+      "loss": 0.0912,
+      "step": 1417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0141960773185625e-05,
+      "loss": 0.0925,
+      "step": 1418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.96825801561629e-05,
+      "loss": 0.0887,
+      "step": 1419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.922484932826899e-05,
+      "loss": 0.0933,
+      "step": 1420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.87687700045379e-05,
+      "loss": 0.0906,
+      "step": 1421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.831434389381546e-05,
+      "loss": 0.0931,
+      "step": 1422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.786157269875386e-05,
+      "loss": 0.0927,
+      "step": 1423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7410458115803865e-05,
+      "loss": 0.0911,
+      "step": 1424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.696100183521002e-05,
+      "loss": 0.0954,
+      "step": 1425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.65132055410027e-05,
+      "loss": 0.0928,
+      "step": 1426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6067070910993335e-05,
+      "loss": 0.0965,
+      "step": 1427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5622599616766914e-05,
+      "loss": 0.0923,
+      "step": 1428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.51797933236764e-05,
+      "loss": 0.0946,
+      "step": 1429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.473865369083631e-05,
+      "loss": 0.0944,
+      "step": 1430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.429918237111642e-05,
+      "loss": 0.0881,
+      "step": 1431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.386138101113569e-05,
+      "loss": 0.0949,
+      "step": 1432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3425251251256024e-05,
+      "loss": 0.0898,
+      "step": 1433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.299079472557622e-05,
+      "loss": 0.0914,
+      "step": 1434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.255801306192559e-05,
+      "loss": 0.0904,
+      "step": 1435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.21269078818582e-05,
+      "loss": 0.0906,
+      "step": 1436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.169748080064651e-05,
+      "loss": 0.0924,
+      "step": 1437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.126973342727587e-05,
+      "loss": 0.0935,
+      "step": 1438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.084366736443724e-05,
+      "loss": 0.0918,
+      "step": 1439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.041928420852299e-05,
+      "loss": 0.0937,
+      "step": 1440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.999658554961917e-05,
+      "loss": 0.0897,
+      "step": 1441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.957557297150056e-05,
+      "loss": 0.0927,
+      "step": 1442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.915624805162488e-05,
+      "loss": 0.0906,
+      "step": 1443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8738612361125864e-05,
+      "loss": 0.093,
+      "step": 1444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8322667464808625e-05,
+      "loss": 0.0909,
+      "step": 1445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7908414921142565e-05,
+      "loss": 0.0921,
+      "step": 1446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7495856282256775e-05,
+      "loss": 0.0905,
+      "step": 1447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.70849930939331e-05,
+      "loss": 0.089,
+      "step": 1448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.667582689560113e-05,
+      "loss": 0.0923,
+      "step": 1449
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.626835922033201e-05,
+      "loss": 0.092,
+      "step": 1450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.586259159483286e-05,
+      "loss": 0.0936,
+      "step": 1451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.545852553944102e-05,
+      "loss": 0.0977,
+      "step": 1452
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.505616256811834e-05,
+      "loss": 0.0922,
+      "step": 1453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4655504188445604e-05,
+      "loss": 0.0911,
+      "step": 1454
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4256551901616704e-05,
+      "loss": 0.0918,
+      "step": 1455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3859307202433144e-05,
+      "loss": 0.0912,
+      "step": 1456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.346377157929837e-05,
+      "loss": 0.0991,
+      "step": 1457
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.306994651421253e-05,
+      "loss": 0.0926,
+      "step": 1458
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.267783348276599e-05,
+      "loss": 0.0948,
+      "step": 1459
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.22874339541352e-05,
+      "loss": 0.0948,
+      "step": 1460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.189874939107574e-05,
+      "loss": 0.092,
+      "step": 1461
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1511781249918e-05,
+      "loss": 0.091,
+      "step": 1462
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.112653098056113e-05,
+      "loss": 0.0896,
+      "step": 1463
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0743000026467414e-05,
+      "loss": 0.0923,
+      "step": 1464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.036118982465786e-05,
+      "loss": 0.0909,
+      "step": 1465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9981101805705246e-05,
+      "loss": 0.0945,
+      "step": 1466
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.960273739373044e-05,
+      "loss": 0.0928,
+      "step": 1467
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.922609800639587e-05,
+      "loss": 0.0903,
+      "step": 1468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.885118505490065e-05,
+      "loss": 0.0927,
+      "step": 1469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8477999943975276e-05,
+      "loss": 0.0904,
+      "step": 1470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.810654407187636e-05,
+      "loss": 0.093,
+      "step": 1471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.773681883038138e-05,
+      "loss": 0.0927,
+      "step": 1472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.736882560478338e-05,
+      "loss": 0.0895,
+      "step": 1473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7002565773886e-05,
+      "loss": 0.0927,
+      "step": 1474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6638040709998046e-05,
+      "loss": 0.0937,
+      "step": 1475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6275251778928484e-05,
+      "loss": 0.0925,
+      "step": 1476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.591420033998122e-05,
+      "loss": 0.0882,
+      "step": 1477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5554887745950446e-05,
+      "loss": 0.0974,
+      "step": 1478
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5197315343114714e-05,
+      "loss": 0.0959,
+      "step": 1479
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4841484471232975e-05,
+      "loss": 0.095,
+      "step": 1480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.448739646353821e-05,
+      "loss": 0.0924,
+      "step": 1481
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.413505264673411e-05,
+      "loss": 0.0914,
+      "step": 1482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.378445434098837e-05,
+      "loss": 0.0923,
+      "step": 1483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3435602859929194e-05,
+      "loss": 0.0916,
+      "step": 1484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.308849951063952e-05,
+      "loss": 0.0938,
+      "step": 1485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2743145593652045e-05,
+      "loss": 0.0943,
+      "step": 1486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2399542402945345e-05,
+      "loss": 0.0859,
+      "step": 1487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2057691225937655e-05,
+      "loss": 0.09,
+      "step": 1488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.171759334348317e-05,
+      "loss": 0.0928,
+      "step": 1489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1379250029866744e-05,
+      "loss": 0.0936,
+      "step": 1490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.104266255279897e-05,
+      "loss": 0.0878,
+      "step": 1491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.070783217341189e-05,
+      "loss": 0.0907,
+      "step": 1492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0374760146253845e-05,
+      "loss": 0.0923,
+      "step": 1493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0043447719285023e-05,
+      "loss": 0.0911,
+      "step": 1494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9713896133872764e-05,
+      "loss": 0.0914,
+      "step": 1495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9386106624786802e-05,
+      "loss": 0.0964,
+      "step": 1496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9060080420194646e-05,
+      "loss": 0.0938,
+      "step": 1497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8735818741657037e-05,
+      "loss": 0.0915,
+      "step": 1498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8413322804123452e-05,
+      "loss": 0.0953,
+      "step": 1499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8092593815927492e-05,
+      "loss": 0.0919,
+      "step": 1500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7773632978781938e-05,
+      "loss": 0.0914,
+      "step": 1501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7456441487775262e-05,
+      "loss": 0.0903,
+      "step": 1502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.714102053136591e-05,
+      "loss": 0.0924,
+      "step": 1503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.682737129137891e-05,
+      "loss": 0.0917,
+      "step": 1504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.651549494300065e-05,
+      "loss": 0.09,
+      "step": 1505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6205392654775116e-05,
+      "loss": 0.089,
+      "step": 1506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5897065588599055e-05,
+      "loss": 0.091,
+      "step": 1507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5590514899717587e-05,
+      "loss": 0.0892,
+      "step": 1508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5285741736720492e-05,
+      "loss": 0.0916,
+      "step": 1509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4982747241537207e-05,
+      "loss": 0.0922,
+      "step": 1510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.468153254943284e-05,
+      "loss": 0.0888,
+      "step": 1511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4382098789004048e-05,
+      "loss": 0.0936,
+      "step": 1512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.40844470821745e-05,
+      "loss": 0.0951,
+      "step": 1513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.378857854419092e-05,
+      "loss": 0.0913,
+      "step": 1514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.349449428361877e-05,
+      "loss": 0.0916,
+      "step": 1515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.32021954023382e-05,
+      "loss": 0.0929,
+      "step": 1516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2911682995539863e-05,
+      "loss": 0.0971,
+      "step": 1517
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2622958151720774e-05,
+      "loss": 0.0936,
+      "step": 1518
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.233602195268025e-05,
+      "loss": 0.0894,
+      "step": 1519
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2050875473516196e-05,
+      "loss": 0.0934,
+      "step": 1520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1767519782620093e-05,
+      "loss": 0.0913,
+      "step": 1521
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1485955941674518e-05,
+      "loss": 0.091,
+      "step": 1522
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1206185005647473e-05,
+      "loss": 0.0903,
+      "step": 1523
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0928208022789942e-05,
+      "loss": 0.0901,
+      "step": 1524
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.06520260346309e-05,
+      "loss": 0.0889,
+      "step": 1525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0377640075973925e-05,
+      "loss": 0.091,
+      "step": 1526
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.010505117489336e-05,
+      "loss": 0.0904,
+      "step": 1527
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9834260352729837e-05,
+      "loss": 0.0911,
+      "step": 1528
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9565268624087475e-05,
+      "loss": 0.0925,
+      "step": 1529
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9298076996829174e-05,
+      "loss": 0.0918,
+      "step": 1530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.903268647207329e-05,
+      "loss": 0.0923,
+      "step": 1531
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8769098044189693e-05,
+      "loss": 0.0906,
+      "step": 1532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8507312700796254e-05,
+      "loss": 0.0869,
+      "step": 1533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8247331422754922e-05,
+      "loss": 0.0914,
+      "step": 1534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.798915518416816e-05,
+      "loss": 0.0942,
+      "step": 1535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7732784952375237e-05,
+      "loss": 0.0928,
+      "step": 1536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7478221687948825e-05,
+      "loss": 0.0901,
+      "step": 1537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7225466344690964e-05,
+      "loss": 0.0916,
+      "step": 1538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.697451986962989e-05,
+      "loss": 0.0901,
+      "step": 1539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6725383203016363e-05,
+      "loss": 0.0946,
+      "step": 1540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6478057278319915e-05,
+      "loss": 0.0907,
+      "step": 1541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6232543022225987e-05,
+      "loss": 0.0927,
+      "step": 1542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5988841354631466e-05,
+      "loss": 0.0948,
+      "step": 1543
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5746953188642333e-05,
+      "loss": 0.0925,
+      "step": 1544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5506879430569275e-05,
+      "loss": 0.0927,
+      "step": 1545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5268620979924986e-05,
+      "loss": 0.0949,
+      "step": 1546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5032178729420476e-05,
+      "loss": 0.0879,
+      "step": 1547
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4797553564961764e-05,
+      "loss": 0.0918,
+      "step": 1548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4564746365646641e-05,
+      "loss": 0.0901,
+      "step": 1549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4333758003761077e-05,
+      "loss": 0.0902,
+      "step": 1550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4104589344776542e-05,
+      "loss": 0.0933,
+      "step": 1551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3877241247346128e-05,
+      "loss": 0.0906,
+      "step": 1552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3651714563301775e-05,
+      "loss": 0.0895,
+      "step": 1553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3428010137650936e-05,
+      "loss": 0.0884,
+      "step": 1554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3206128808573193e-05,
+      "loss": 0.0924,
+      "step": 1555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2986071407417532e-05,
+      "loss": 0.0912,
+      "step": 1556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2767838758698912e-05,
+      "loss": 0.0938,
+      "step": 1557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.255143168009526e-05,
+      "loss": 0.0892,
+      "step": 1558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2336850982444525e-05,
+      "loss": 0.0936,
+      "step": 1559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2124097469741469e-05,
+      "loss": 0.091,
+      "step": 1560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.191317193913466e-05,
+      "loss": 0.0902,
+      "step": 1561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1704075180923756e-05,
+      "loss": 0.091,
+      "step": 1562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1496807978556012e-05,
+      "loss": 0.094,
+      "step": 1563
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1291371108624104e-05,
+      "loss": 0.0931,
+      "step": 1564
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1087765340862311e-05,
+      "loss": 0.0949,
+      "step": 1565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0885991438144449e-05,
+      "loss": 0.0931,
+      "step": 1566
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0686050156480608e-05,
+      "loss": 0.09,
+      "step": 1567
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.048794224501426e-05,
+      "loss": 0.0884,
+      "step": 1568
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0291668446019641e-05,
+      "loss": 0.0941,
+      "step": 1569
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0097229494898996e-05,
+      "loss": 0.0969,
+      "step": 1570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.904626120179506e-06,
+      "loss": 0.0906,
+      "step": 1571
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.71385904351102e-06,
+      "loss": 0.0927,
+      "step": 1572
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.524928979662951e-06,
+      "loss": 0.0904,
+      "step": 1573
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.337836636521879e-06,
+      "loss": 0.0916,
+      "step": 1574
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.152582715088675e-06,
+      "loss": 0.0957,
+      "step": 1575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.969167909475938e-06,
+      "loss": 0.0937,
+      "step": 1576
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.787592906905729e-06,
+      "loss": 0.0876,
+      "step": 1577
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.607858387706336e-06,
+      "loss": 0.0976,
+      "step": 1578
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.429965025310293e-06,
+      "loss": 0.0907,
+      "step": 1579
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.253913486251542e-06,
+      "loss": 0.0913,
+      "step": 1580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.079704430163205e-06,
+      "loss": 0.0906,
+      "step": 1581
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.907338509774875e-06,
+      "loss": 0.0945,
+      "step": 1582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.736816370910116e-06,
+      "loss": 0.0925,
+      "step": 1583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.5681386524845135e-06,
+      "loss": 0.0963,
+      "step": 1584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.401305986502516e-06,
+      "loss": 0.0909,
+      "step": 1585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.23631899805588e-06,
+      "loss": 0.0897,
+      "step": 1586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.073178305320726e-06,
+      "loss": 0.0957,
+      "step": 1587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.911884519555489e-06,
+      "loss": 0.0939,
+      "step": 1588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.75243824509858e-06,
+      "loss": 0.0888,
+      "step": 1589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.594840079366171e-06,
+      "loss": 0.0937,
+      "step": 1590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.439090612849863e-06,
+      "loss": 0.0915,
+      "step": 1591
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.2851904291145175e-06,
+      "loss": 0.093,
+      "step": 1592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.133140104796098e-06,
+      "loss": 0.0919,
+      "step": 1593
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9829402095993855e-06,
+      "loss": 0.0928,
+      "step": 1594
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.834591306296044e-06,
+      "loss": 0.091,
+      "step": 1595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.688093950722451e-06,
+      "loss": 0.0929,
+      "step": 1596
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.543448691777419e-06,
+      "loss": 0.0937,
+      "step": 1597
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.400656071420373e-06,
+      "loss": 0.0905,
+      "step": 1598
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2597166246693415e-06,
+      "loss": 0.0937,
+      "step": 1599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.120630879598631e-06,
+      "loss": 0.0923,
+      "step": 1600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.983399357337215e-06,
+      "loss": 0.0949,
+      "step": 1601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8480225720665685e-06,
+      "loss": 0.0905,
+      "step": 1602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.714501031018892e-06,
+      "loss": 0.093,
+      "step": 1603
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.582835234475114e-06,
+      "loss": 0.0909,
+      "step": 1604
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.453025675762834e-06,
+      "loss": 0.0928,
+      "step": 1605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.32507284125494e-06,
+      "loss": 0.0963,
+      "step": 1606
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.198977210367272e-06,
+      "loss": 0.0895,
+      "step": 1607
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.074739255557292e-06,
+      "loss": 0.0886,
+      "step": 1608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.952359442321862e-06,
+      "loss": 0.0966,
+      "step": 1609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.831838229195972e-06,
+      "loss": 0.0929,
+      "step": 1610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7131760677505677e-06,
+      "loss": 0.0948,
+      "step": 1611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5963734025911132e-06,
+      "loss": 0.0894,
+      "step": 1612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4814306713559785e-06,
+      "loss": 0.0923,
+      "step": 1613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.368348304714608e-06,
+      "loss": 0.0901,
+      "step": 1614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2571267263659664e-06,
+      "loss": 0.0919,
+      "step": 1615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1477663530371513e-06,
+      "loss": 0.0931,
+      "step": 1616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0402675944814496e-06,
+      "loss": 0.0913,
+      "step": 1617
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.934630853477116e-06,
+      "loss": 0.0923,
+      "step": 1618
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8308565258258203e-06,
+      "loss": 0.0911,
+      "step": 1619
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.728945000351035e-06,
+      "loss": 0.0941,
+      "step": 1620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.628896658896762e-06,
+      "loss": 0.0914,
+      "step": 1621
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5307118763257533e-06,
+      "loss": 0.0936,
+      "step": 1622
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4343910205185116e-06,
+      "loss": 0.0925,
+      "step": 1623
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3399344523719058e-06,
+      "loss": 0.0936,
+      "step": 1624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2473425257972245e-06,
+      "loss": 0.0923,
+      "step": 1625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.15661558771979e-06,
+      "loss": 0.0879,
+      "step": 1626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.067753978076736e-06,
+      "loss": 0.0938,
+      "step": 1627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9807580298163986e-06,
+      "loss": 0.0897,
+      "step": 1628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8956280688967043e-06,
+      "loss": 0.093,
+      "step": 1629
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8123644142841178e-06,
+      "loss": 0.0936,
+      "step": 1630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7309673779524193e-06,
+      "loss": 0.0938,
+      "step": 1631
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6514372648813726e-06,
+      "loss": 0.0928,
+      "step": 1632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5737743730559473e-06,
+      "loss": 0.0898,
+      "step": 1633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4979789934647103e-06,
+      "loss": 0.0886,
+      "step": 1634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4240514100992142e-06,
+      "loss": 0.0923,
+      "step": 1635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3519918999526648e-06,
+      "loss": 0.091,
+      "step": 1636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.281800733018812e-06,
+      "loss": 0.0926,
+      "step": 1637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2134781722912825e-06,
+      "loss": 0.0944,
+      "step": 1638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1470244737621926e-06,
+      "loss": 0.0926,
+      "step": 1639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.082439886421427e-06,
+      "loss": 0.0877,
+      "step": 1640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0197246522557491e-06,
+      "loss": 0.0915,
+      "step": 1641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.588790062476372e-07,
+      "loss": 0.0908,
+      "step": 1642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.999031763746169e-07,
+      "loss": 0.0953,
+      "step": 1643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.427973836084846e-07,
+      "loss": 0.0932,
+      "step": 1644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.875618419141417e-07,
+      "loss": 0.0951,
+      "step": 1645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.341967582490949e-07,
+      "loss": 0.0938,
+      "step": 1646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.827023325626236e-07,
+      "loss": 0.0912,
+      "step": 1647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.330787577950026e-07,
+      "loss": 0.0953,
+      "step": 1648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.85326219876614e-07,
+      "loss": 0.0925,
+      "step": 1649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.394448977276145e-07,
+      "loss": 0.0907,
+      "step": 1650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.954349632569355e-07,
+      "loss": 0.0918,
+      "step": 1651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.532965813617285e-07,
+      "loss": 0.0899,
+      "step": 1652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.130299099268653e-07,
+      "loss": 0.0913,
+      "step": 1653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.746350998242165e-07,
+      "loss": 0.0897,
+      "step": 1654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.381122949121518e-07,
+      "loss": 0.0923,
+      "step": 1655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.034616320349293e-07,
+      "loss": 0.0886,
+      "step": 1656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.70683241022307e-07,
+      "loss": 0.0899,
+      "step": 1657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.397772446889879e-07,
+      "loss": 0.0938,
+      "step": 1658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1074375883400888e-07,
+      "loss": 0.0938,
+      "step": 1659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8358289224057468e-07,
+      "loss": 0.0957,
+      "step": 1660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.58294746675558e-07,
+      "loss": 0.0904,
+      "step": 1661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3487941688899996e-07,
+      "loss": 0.0877,
+      "step": 1662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1333699061388814e-07,
+      "loss": 0.0934,
+      "step": 1663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.366754856571236e-08,
+      "loss": 0.0906,
+      "step": 1664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.587116444235376e-08,
+      "loss": 0.0913,
+      "step": 1665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.994790492352964e-08,
+      "loss": 0.0942,
+      "step": 1666
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.58978296709045e-08,
+      "loss": 0.0887,
+      "step": 1667
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3720991327534924e-08,
+      "loss": 0.091,
+      "step": 1668
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3417435517703034e-08,
+      "loss": 0.094,
+      "step": 1669
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.498720084708305e-08,
+      "loss": 0.0945,
+      "step": 1670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.430318902130641e-09,
+      "loss": 0.0897,
+      "step": 1671
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.746814250360498e-09,
+      "loss": 0.091,
+      "step": 1672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.367044400132585e-10,
+      "loss": 0.0913,
+      "step": 1673
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.0884,
+      "step": 1674
+    },
+    {
+      "epoch": 1.0,
+      "step": 1674,
+      "total_flos": 333731052404736.0,
+      "train_loss": 0.13337620469689226,
+      "train_runtime": 22495.1749,
+      "train_samples_per_second": 9.53,
+      "train_steps_per_second": 0.074
+    }
+  ],
+  "max_steps": 1674,
+  "num_train_epochs": 1,
+  "total_flos": 333731052404736.0,
+  "trial_name": null,
+  "trial_params": null
+}