diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7621 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5001481481481481,
+  "global_step": 1266,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.631578947368421e-05,
+      "loss": 6.8257,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.263157894736842e-05,
+      "loss": 6.7129,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.894736842105263e-05,
+      "loss": 5.1377,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010526315789473683,
+      "loss": 6.9224,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013157894736842105,
+      "loss": 4.7139,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015789473684210527,
+      "loss": 4.3567,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018421052631578948,
+      "loss": 3.8765,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00021052631578947367,
+      "loss": 3.5278,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00023684210526315788,
+      "loss": 3.2664,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002631578947368421,
+      "loss": 3.05,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00028947368421052634,
+      "loss": 2.8108,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00031578947368421053,
+      "loss": 2.564,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034210526315789477,
+      "loss": 2.2808,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036842105263157896,
+      "loss": 1.9471,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039473684210526315,
+      "loss": 1.5404,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042105263157894734,
+      "loss": 1.2093,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004473684210526316,
+      "loss": 0.9499,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047368421052631577,
+      "loss": 0.7274,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005,
+      "loss": 0.6061,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005263157894736842,
+      "loss": 0.4458,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005526315789473685,
+      "loss": 0.4048,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0005789473684210527,
+      "loss": 0.341,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006052631578947369,
+      "loss": 0.2969,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006315789473684211,
+      "loss": 0.2816,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006578947368421054,
+      "loss": 0.2455,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0006842105263157895,
+      "loss": 0.2198,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007105263157894737,
+      "loss": 0.228,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007368421052631579,
+      "loss": 0.2194,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007631578947368421,
+      "loss": 0.2119,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0007894736842105263,
+      "loss": 0.1943,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008157894736842105,
+      "loss": 0.1907,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008421052631578947,
+      "loss": 0.1796,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000868421052631579,
+      "loss": 0.1802,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0008947368421052632,
+      "loss": 0.1722,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009210526315789473,
+      "loss": 0.1831,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009473684210526315,
+      "loss": 0.1684,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0009736842105263158,
+      "loss": 0.1716,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.001,
+      "loss": 0.1535,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009999983637759059,
+      "loss": 0.1592,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009999934551143319,
+      "loss": 0.1601,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009999852740474053,
+      "loss": 0.1554,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00099997382062867,
+      "loss": 0.1487,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009999590949330874,
+      "loss": 0.1512,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009999410970570357,
+      "loss": 0.1544,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009999198271183093,
+      "loss": 0.1536,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009998952852561175,
+      "loss": 0.147,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009998674716310844,
+      "loss": 0.1447,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009998363864252475,
+      "loss": 0.1411,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009998020298420558,
+      "loss": 0.1398,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009997644021063697,
+      "loss": 0.1424,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000999723503464459,
+      "loss": 0.146,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000999679334184001,
+      "loss": 0.1471,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009996318945540792,
+      "loss": 0.1295,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009995811848851806,
+      "loss": 0.132,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009995272055091952,
+      "loss": 0.1333,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009994699567794124,
+      "loss": 0.1283,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009994094390705188,
+      "loss": 0.1257,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000999345652778597,
+      "loss": 0.1365,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009992785983211215,
+      "loss": 0.1287,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009992082761369567,
+      "loss": 0.1348,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000999134686686354,
+      "loss": 0.1203,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009990578304509487,
+      "loss": 0.1228,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0009989777079337573,
+      "loss": 0.1284,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009988943196591726,
+      "loss": 0.1243,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009988076661729631,
+      "loss": 0.1357,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009987177480422662,
+      "loss": 0.1245,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000998624565855587,
+      "loss": 0.1304,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009985281202227936,
+      "loss": 0.1271,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000998428411775112,
+      "loss": 0.127,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009983254411651242,
+      "loss": 0.1278,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009982192090667618,
+      "loss": 0.1189,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009981097161753032,
+      "loss": 0.1242,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009979969632073677,
+      "loss": 0.1201,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000997880950900912,
+      "loss": 0.1252,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009977616800152247,
+      "loss": 0.1203,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009976391513309211,
+      "loss": 0.1218,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009975133656499391,
+      "loss": 0.1215,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009973843237955327,
+      "loss": 0.1222,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009972520266122676,
+      "loss": 0.1175,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009971164749660148,
+      "loss": 0.1271,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009969776697439463,
+      "loss": 0.1159,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009968356118545277,
+      "loss": 0.1152,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000996690302227513,
+      "loss": 0.1174,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000996541741813939,
+      "loss": 0.1162,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009963899315861176,
+      "loss": 0.1191,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009962348725376317,
+      "loss": 0.1146,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009960765656833262,
+      "loss": 0.1181,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009959150120593034,
+      "loss": 0.1126,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000995750212722915,
+      "loss": 0.1149,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009955821687527553,
+      "loss": 0.1176,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000995410881248655,
+      "loss": 0.1178,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009952363513316726,
+      "loss": 0.1151,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009950585801440889,
+      "loss": 0.1144,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009948775688493974,
+      "loss": 0.1144,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009946933186322986,
+      "loss": 0.1159,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000994505830698691,
+      "loss": 0.1164,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009943151062756638,
+      "loss": 0.1117,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009941211466114883,
+      "loss": 0.1121,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009939239529756106,
+      "loss": 0.1183,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009937235266586424,
+      "loss": 0.1152,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009935198689723535,
+      "loss": 0.1113,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009933129812496624,
+      "loss": 0.1165,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009931028648446274,
+      "loss": 0.1134,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009928895211324387,
+      "loss": 0.1146,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000992672951509409,
+      "loss": 0.1132,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000992453157392964,
+      "loss": 0.1098,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009922301402216333,
+      "loss": 0.11,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009920039014550412,
+      "loss": 0.1112,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000991774442573897,
+      "loss": 0.1079,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009915417650799853,
+      "loss": 0.1116,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009913058704961563,
+      "loss": 0.1118,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009910667603663156,
+      "loss": 0.1115,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000990824436255414,
+      "loss": 0.1084,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009905788997494377,
+      "loss": 0.11,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000990330152455398,
+      "loss": 0.1103,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00099007819600132,
+      "loss": 0.1142,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009898230320362323,
+      "loss": 0.1115,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000989564662230157,
+      "loss": 0.117,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009893030882740976,
+      "loss": 0.1084,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009890383118800286,
+      "loss": 0.1113,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000988770334780884,
+      "loss": 0.1116,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000988499158730546,
+      "loss": 0.1147,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009882247855038339,
+      "loss": 0.1066,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000987947216896492,
+      "loss": 0.1084,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000987666454725178,
+      "loss": 0.1092,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009873825008274513,
+      "loss": 0.1103,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000987095357061761,
+      "loss": 0.113,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009868050253074327,
+      "loss": 0.1083,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009865115074646582,
+      "loss": 0.1078,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009862148054544811,
+      "loss": 0.1085,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009859149212187853,
+      "loss": 0.109,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009856118567202821,
+      "loss": 0.1084,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009853056139424974,
+      "loss": 0.1116,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009849961948897581,
+      "loss": 0.1085,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009846836015871801,
+      "loss": 0.1074,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000984367836080654,
+      "loss": 0.1072,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009840489004368325,
+      "loss": 0.1112,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009837267967431162,
+      "loss": 0.1107,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009834015271076406,
+      "loss": 0.1072,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009830730936592614,
+      "loss": 0.1097,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009827414985475419,
+      "loss": 0.1095,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009824067439427374,
+      "loss": 0.1099,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009820688320357823,
+      "loss": 0.1068,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000981727765038275,
+      "loss": 0.1076,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009813835451824636,
+      "loss": 0.1095,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009810361747212312,
+      "loss": 0.1071,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009806856559280818,
+      "loss": 0.1103,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009803319910971246,
+      "loss": 0.107,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009799751825430591,
+      "loss": 0.1131,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009796152326011602,
+      "loss": 0.1045,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009792521436272632,
+      "loss": 0.1093,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009788859179977478,
+      "loss": 0.1087,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009785165581095226,
+      "loss": 0.1068,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009781440663800098,
+      "loss": 0.1087,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009777684452471295,
+      "loss": 0.1087,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009773896971692828,
+      "loss": 0.1051,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009770078246253366,
+      "loss": 0.1116,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009766228301146074,
+      "loss": 0.1111,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009762347161568441,
+      "loss": 0.1071,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009758434852922123,
+      "loss": 0.1091,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009754491400812778,
+      "loss": 0.1111,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009750516831049889,
+      "loss": 0.1074,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009746511169646604,
+      "loss": 0.1092,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009742474442819561,
+      "loss": 0.1123,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000973840667698872,
+      "loss": 0.1058,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009734307898777186,
+      "loss": 0.1068,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009730178135011037,
+      "loss": 0.1113,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009726017412719151,
+      "loss": 0.1078,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009721825759133022,
+      "loss": 0.1079,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009717603201686588,
+      "loss": 0.1097,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009713349768016054,
+      "loss": 0.1093,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009709065485959699,
+      "loss": 0.1092,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009704750383557705,
+      "loss": 0.1108,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009700404489051972,
+      "loss": 0.1091,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000969602783088593,
+      "loss": 0.1073,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000969162043770435,
+      "loss": 0.1072,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009687182338353165,
+      "loss": 0.1087,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009682713561879275,
+      "loss": 0.1101,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000967821413753036,
+      "loss": 0.109,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009673684094754685,
+      "loss": 0.1084,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009669123463200913,
+      "loss": 0.1074,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009664532272717901,
+      "loss": 0.1085,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009659910553354519,
+      "loss": 0.1083,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009655258335359438,
+      "loss": 0.1064,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009650575649180944,
+      "loss": 0.1065,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009645862525466733,
+      "loss": 0.1076,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009641118995063711,
+      "loss": 0.1056,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009636345089017794,
+      "loss": 0.1095,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00096315408385737,
+      "loss": 0.1088,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009626706275174754,
+      "loss": 0.1064,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009621841430462669,
+      "loss": 0.1065,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000961694633627735,
+      "loss": 0.1061,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009612021024656684,
+      "loss": 0.1097,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009607065527836323,
+      "loss": 0.1095,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009602079878249478,
+      "loss": 0.1086,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009597064108526714,
+      "loss": 0.1069,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000959201825149572,
+      "loss": 0.1066,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009586942340181109,
+      "loss": 0.1078,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009581836407804195,
+      "loss": 0.1062,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009576700487782774,
+      "loss": 0.1078,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009571534613730914,
+      "loss": 0.1054,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009566338819458724,
+      "loss": 0.1082,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009561113138972137,
+      "loss": 0.1094,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009555857606472692,
+      "loss": 0.1073,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009550572256357304,
+      "loss": 0.1043,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009545257123218043,
+      "loss": 0.1122,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009539912241841904,
+      "loss": 0.1072,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009534537647210581,
+      "loss": 0.106,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000952913337450024,
+      "loss": 0.1096,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009523699459081285,
+      "loss": 0.1124,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009518235936518129,
+      "loss": 0.1061,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009512742842568963,
+      "loss": 0.1081,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009507220213185517,
+      "loss": 0.1098,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009501668084512827,
+      "loss": 0.1083,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009496086492889,
+      "loss": 0.1065,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009490475474844975,
+      "loss": 0.1095,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009484835067104284,
+      "loss": 0.107,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009479165306582811,
+      "loss": 0.1089,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000947346623038855,
+      "loss": 0.1093,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009467737875821367,
+      "loss": 0.1065,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009461980280372747,
+      "loss": 0.1056,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009456193481725555,
+      "loss": 0.1073,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000945037751775379,
+      "loss": 0.107,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009444532426522334,
+      "loss": 0.1115,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00094386582462867,
+      "loss": 0.1075,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009432755015492794,
+      "loss": 0.1062,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009426822772776644,
+      "loss": 0.1072,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009420861556964169,
+      "loss": 0.1073,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009414871407070906,
+      "loss": 0.1044,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009408852362301767,
+      "loss": 0.1059,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009402804462050776,
+      "loss": 0.1053,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000939672774590081,
+      "loss": 0.1062,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009390622253623352,
+      "loss": 0.1072,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009384488025178212,
+      "loss": 0.1064,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009378325100713282,
+      "loss": 0.1062,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009372133520564263,
+      "loss": 0.1103,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009365913325254406,
+      "loss": 0.1087,
+      "step": 237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009359664555494242,
+      "loss": 0.1049,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009353387252181327,
+      "loss": 0.1085,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009347081456399957,
+      "loss": 0.1049,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009340747209420912,
+      "loss": 0.1098,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009334384552701182,
+      "loss": 0.1055,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009327993527883698,
+      "loss": 0.1072,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009321574176797054,
+      "loss": 0.1091,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009315126541455237,
+      "loss": 0.1033,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009308650664057352,
+      "loss": 0.1056,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009302146586987347,
+      "loss": 0.111,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009295614352813731,
+      "loss": 0.1073,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009289054004289301,
+      "loss": 0.106,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009282465584350857,
+      "loss": 0.1061,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009275849136118926,
+      "loss": 0.108,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009269204702897476,
+      "loss": 0.1085,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009262532328173632,
+      "loss": 0.106,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009255832055617398,
+      "loss": 0.1069,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009249103929081362,
+      "loss": 0.1065,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009242347992600415,
+      "loss": 0.1069,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009235564290391461,
+      "loss": 0.107,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009228752866853129,
+      "loss": 0.1064,
+      "step": 258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009221913766565479,
+      "loss": 0.107,
+      "step": 259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009215047034289715,
+      "loss": 0.1068,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009208152714967888,
+      "loss": 0.103,
+      "step": 261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009201230853722602,
+      "loss": 0.1042,
+      "step": 262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009194281495856722,
+      "loss": 0.1076,
+      "step": 263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009187304686853078,
+      "loss": 0.1047,
+      "step": 264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009180300472374158,
+      "loss": 0.1061,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009173268898261821,
+      "loss": 0.1063,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009166210010536995,
+      "loss": 0.1061,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009159123855399363,
+      "loss": 0.1062,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009152010479227077,
+      "loss": 0.1065,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009144869928576451,
+      "loss": 0.1058,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009137702250181645,
+      "loss": 0.1078,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009130507490954374,
+      "loss": 0.103,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000912328569798359,
+      "loss": 0.1037,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000911603691853518,
+      "loss": 0.1074,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009108761200051654,
+      "loss": 0.1069,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009101458590151836,
+      "loss": 0.1083,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009094129136630551,
+      "loss": 0.1082,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009086772887458314,
+      "loss": 0.1076,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009079389890781011,
+      "loss": 0.1042,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009071980194919592,
+      "loss": 0.1046,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009064543848369748,
+      "loss": 0.1053,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009057080899801597,
+      "loss": 0.1036,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009049591398059362,
+      "loss": 0.1079,
+      "step": 283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009042075392161061,
+      "loss": 0.1048,
+      "step": 284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009034532931298169,
+      "loss": 0.1051,
+      "step": 285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009026964064835311,
+      "loss": 0.1044,
+      "step": 286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009019368842309936,
+      "loss": 0.1085,
+      "step": 287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009011747313431988,
+      "loss": 0.1037,
+      "step": 288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009004099528083583,
+      "loss": 0.1039,
+      "step": 289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008996425536318682,
+      "loss": 0.1071,
+      "step": 290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0008988725388362768,
+      "loss": 0.1062,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000898099913461251,
+      "loss": 0.1068,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008973246825635441,
+      "loss": 0.1048,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008965468512169618,
+      "loss": 0.1049,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008957664245123296,
+      "loss": 0.1046,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008949834075574596,
+      "loss": 0.1074,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008941978054771163,
+      "loss": 0.1059,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008934096234129843,
+      "loss": 0.1055,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008926188665236333,
+      "loss": 0.107,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008918255399844854,
+      "loss": 0.1041,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008910296489877803,
+      "loss": 0.1057,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008902311987425422,
+      "loss": 0.1088,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008894301944745453,
+      "loss": 0.1044,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008886266414262796,
+      "loss": 0.104,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008878205448569163,
+      "loss": 0.1034,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008870119100422742,
+      "loss": 0.1059,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008862007422747842,
+      "loss": 0.1047,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008853870468634552,
+      "loss": 0.1046,
+      "step": 308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008845708291338396,
+      "loss": 0.1043,
+      "step": 309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008837520944279976,
+      "loss": 0.1075,
+      "step": 310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008829308481044632,
+      "loss": 0.1056,
+      "step": 311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008821070955382082,
+      "loss": 0.1021,
+      "step": 312
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008812808421206082,
+      "loss": 0.1057,
+      "step": 313
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000880452093259406,
+      "loss": 0.1078,
+      "step": 314
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000879620854378677,
+      "loss": 0.1046,
+      "step": 315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0008787871309187936,
+      "loss": 0.1043,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008779509283363894,
+      "loss": 0.1081,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008771122521043234,
+      "loss": 0.1038,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008762711077116452,
+      "loss": 0.1036,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008754275006635573,
+      "loss": 0.1079,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008745814364813805,
+      "loss": 0.1051,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000873732920702517,
+      "loss": 0.1047,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000872881958880415,
+      "loss": 0.1055,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008720285565845312,
+      "loss": 0.1041,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008711727194002954,
+      "loss": 0.1036,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008703144529290732,
+      "loss": 0.1077,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008694537627881296,
+      "loss": 0.1074,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008685906546105925,
+      "loss": 0.1022,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008677251340454155,
+      "loss": 0.1079,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008668572067573408,
+      "loss": 0.1054,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008659868784268628,
+      "loss": 0.1035,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008651141547501903,
+      "loss": 0.1071,
+      "step": 332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008642390414392091,
+      "loss": 0.1041,
+      "step": 333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008633615442214452,
+      "loss": 0.107,
+      "step": 334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000862481668840027,
+      "loss": 0.1081,
+      "step": 335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008615994210536478,
+      "loss": 0.1038,
+      "step": 336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008607148066365278,
+      "loss": 0.1074,
+      "step": 337
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008598278313783765,
+      "loss": 0.1056,
+      "step": 338
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008589385010843556,
+      "loss": 0.1057,
+      "step": 339
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008580468215750392,
+      "loss": 0.1049,
+      "step": 340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0008571527986863775,
+      "loss": 0.1055,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008562564382696577,
+      "loss": 0.1065,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008553577461914658,
+      "loss": 0.1047,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008544567283336485,
+      "loss": 0.1046,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008535533905932737,
+      "loss": 0.102,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008526477388825941,
+      "loss": 0.1057,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008517397791290058,
+      "loss": 0.1101,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008508295172750116,
+      "loss": 0.1065,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008499169592781806,
+      "loss": 0.1056,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008490021111111108,
+      "loss": 0.1095,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008480849787613883,
+      "loss": 0.1024,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008471655682315495,
+      "loss": 0.106,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008462438855390409,
+      "loss": 0.1056,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008453199367161804,
+      "loss": 0.1053,
+      "step": 354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000844393727810117,
+      "loss": 0.103,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008434652648827924,
+      "loss": 0.1069,
+      "step": 356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008425345540108999,
+      "loss": 0.1055,
+      "step": 357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008416016012858461,
+      "loss": 0.1058,
+      "step": 358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008406664128137094,
+      "loss": 0.1049,
+      "step": 359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000839728994715202,
+      "loss": 0.1019,
+      "step": 360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008387893531256277,
+      "loss": 0.1061,
+      "step": 361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008378474941948438,
+      "loss": 0.1039,
+      "step": 362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008369034240872189,
+      "loss": 0.1034,
+      "step": 363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008359571489815944,
+      "loss": 0.1034,
+      "step": 364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008350086750712429,
+      "loss": 0.1058,
+      "step": 365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008340580085638274,
+      "loss": 0.1026,
+      "step": 366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0008331051556813622,
+      "loss": 0.1033,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008321501226601702,
+      "loss": 0.1053,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000831192915750844,
+      "loss": 0.1056,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008302335412182034,
+      "loss": 0.1072,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008292720053412553,
+      "loss": 0.1059,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008283083144131522,
+      "loss": 0.1057,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008273424747411518,
+      "loss": 0.1034,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008263744926465744,
+      "loss": 0.103,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008254043744647624,
+      "loss": 0.1053,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000824432126545039,
+      "loss": 0.1036,
+      "step": 376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008234577552506661,
+      "loss": 0.1057,
+      "step": 377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008224812669588026,
+      "loss": 0.1033,
+      "step": 378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008215026680604636,
+      "loss": 0.1058,
+      "step": 379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000820521964960477,
+      "loss": 0.1013,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008195391640774432,
+      "loss": 0.1064,
+      "step": 381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008185542718436923,
+      "loss": 0.1011,
+      "step": 382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008175672947052415,
+      "loss": 0.1057,
+      "step": 383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008165782391217543,
+      "loss": 0.1038,
+      "step": 384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008155871115664968,
+      "loss": 0.1052,
+      "step": 385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008145939185262963,
+      "loss": 0.1035,
+      "step": 386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008135986665014981,
+      "loss": 0.1065,
+      "step": 387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008126013620059236,
+      "loss": 0.1025,
+      "step": 388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008116020115668277,
+      "loss": 0.1028,
+      "step": 389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008106006217248552,
+      "loss": 0.103,
+      "step": 390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0008095971990339987,
+      "loss": 0.1041,
+      "step": 391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000808591750061556,
+      "loss": 0.1044,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008075842813880865,
+      "loss": 0.1078,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008065747996073681,
+      "loss": 0.1055,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008055633113263543,
+      "loss": 0.1058,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008045498231651313,
+      "loss": 0.1026,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008035343417568742,
+      "loss": 0.104,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008025168737478034,
+      "loss": 0.108,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0008014974257971414,
+      "loss": 0.1052,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00080047600457707,
+      "loss": 0.1031,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007994526167726848,
+      "loss": 0.1055,
+      "step": 401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000798427269081953,
+      "loss": 0.1063,
+      "step": 402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007973999682156688,
+      "loss": 0.1051,
+      "step": 403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007963707208974103,
+      "loss": 0.1036,
+      "step": 404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007953395338634944,
+      "loss": 0.1069,
+      "step": 405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007943064138629331,
+      "loss": 0.1056,
+      "step": 406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007932713676573901,
+      "loss": 0.1062,
+      "step": 407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007922344020211356,
+      "loss": 0.1039,
+      "step": 408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000791195523741002,
+      "loss": 0.1044,
+      "step": 409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00079015473961634,
+      "loss": 0.1068,
+      "step": 410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000789112056458974,
+      "loss": 0.1047,
+      "step": 411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007880674810931572,
+      "loss": 0.1039,
+      "step": 412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000787021020355527,
+      "loss": 0.1046,
+      "step": 413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007859726810950605,
+      "loss": 0.1027,
+      "step": 414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007849224701730296,
+      "loss": 0.1021,
+      "step": 415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007838703944629559,
+      "loss": 0.1027,
+      "step": 416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0007828164608505661,
+      "loss": 0.1017,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007817606762337464,
+      "loss": 0.1066,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007807030475224974,
+      "loss": 0.1046,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007796435816388898,
+      "loss": 0.104,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007785822855170179,
+      "loss": 0.1022,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000777519166102955,
+      "loss": 0.1038,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007764542303547071,
+      "loss": 0.1037,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007753874852421685,
+      "loss": 0.1033,
+      "step": 424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007743189377470756,
+      "loss": 0.1039,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007732485948629609,
+      "loss": 0.1062,
+      "step": 426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007721764635951076,
+      "loss": 0.1024,
+      "step": 427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007711025509605041,
+      "loss": 0.1053,
+      "step": 428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007700268639877966,
+      "loss": 0.1045,
+      "step": 429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007689494097172456,
+      "loss": 0.1045,
+      "step": 430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007678701952006773,
+      "loss": 0.1049,
+      "step": 431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007667892275014388,
+      "loss": 0.1042,
+      "step": 432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007657065136943519,
+      "loss": 0.101,
+      "step": 433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007646220608656662,
+      "loss": 0.1065,
+      "step": 434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007635358761130132,
+      "loss": 0.1056,
+      "step": 435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007624479665453592,
+      "loss": 0.1058,
+      "step": 436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00076135833928296,
+      "loss": 0.1035,
+      "step": 437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007602670014573128,
+      "loss": 0.1075,
+      "step": 438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007591739602111107,
+      "loss": 0.1081,
+      "step": 439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007580792226981954,
+      "loss": 0.127,
+      "step": 440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007569827960835106,
+      "loss": 0.5453,
+      "step": 441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0007558846875430547,
+      "loss": 0.3859,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007547849042638345,
+      "loss": 0.2602,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007536834534438174,
+      "loss": 0.212,
+      "step": 444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007525803422918849,
+      "loss": 0.253,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007514755780277853,
+      "loss": 0.2305,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007503691678820861,
+      "loss": 0.2013,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007492611190961271,
+      "loss": 0.1775,
+      "step": 448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000748151438921973,
+      "loss": 0.1647,
+      "step": 449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007470401346223653,
+      "loss": 0.1625,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007459272134706755,
+      "loss": 0.1697,
+      "step": 451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007448126827508572,
+      "loss": 0.1674,
+      "step": 452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007436965497573985,
+      "loss": 0.1521,
+      "step": 453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007425788217952743,
+      "loss": 0.1562,
+      "step": 454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007414595061798982,
+      "loss": 0.1554,
+      "step": 455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007403386102370751,
+      "loss": 0.1524,
+      "step": 456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007392161413029523,
+      "loss": 0.1506,
+      "step": 457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007380921067239731,
+      "loss": 0.1393,
+      "step": 458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007369665138568272,
+      "loss": 0.1514,
+      "step": 459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007358393700684033,
+      "loss": 0.1398,
+      "step": 460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007347106827357407,
+      "loss": 0.1294,
+      "step": 461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000733580459245981,
+      "loss": 0.1344,
+      "step": 462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00073244870699632,
+      "loss": 0.1288,
+      "step": 463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007313154333939587,
+      "loss": 0.1288,
+      "step": 464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007301806458560552,
+      "loss": 0.133,
+      "step": 465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0007290443518096769,
+      "loss": 0.1225,
+      "step": 466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00072790655869175,
+      "loss": 0.1237,
+      "step": 467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000726767273949013,
+      "loss": 0.1245,
+      "step": 468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007256265050379665,
+      "loss": 0.1262,
+      "step": 469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007244842594248244,
+      "loss": 0.1201,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000723340544585466,
+      "loss": 0.1183,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007221953680053866,
+      "loss": 0.1171,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007210487371796481,
+      "loss": 0.1187,
+      "step": 473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007199006596128307,
+      "loss": 0.1238,
+      "step": 474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007187511428189828,
+      "loss": 0.1163,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007176001943215728,
+      "loss": 0.1215,
+      "step": 476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007164478216534393,
+      "loss": 0.1149,
+      "step": 477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007152940323567423,
+      "loss": 0.114,
+      "step": 478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007141388339829127,
+      "loss": 0.1159,
+      "step": 479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007129822340926045,
+      "loss": 0.1127,
+      "step": 480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007118242402556438,
+      "loss": 0.1165,
+      "step": 481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007106648600509808,
+      "loss": 0.1135,
+      "step": 482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007095041010666385,
+      "loss": 0.1132,
+      "step": 483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000708341970899664,
+      "loss": 0.1116,
+      "step": 484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000707178477156079,
+      "loss": 0.1077,
+      "step": 485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007060136274508295,
+      "loss": 0.1114,
+      "step": 486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007048474294077361,
+      "loss": 0.1134,
+      "step": 487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007036798906594441,
+      "loss": 0.1105,
+      "step": 488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007025110188473739,
+      "loss": 0.11,
+      "step": 489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0007013408216216698,
+      "loss": 0.1125,
+      "step": 490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000700169306641152,
+      "loss": 0.1102,
+      "step": 491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006989964815732642,
+      "loss": 0.1079,
+      "step": 492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0006978223540940253,
+      "loss": 0.107,
+      "step": 493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006966469318879777,
+      "loss": 0.1069,
+      "step": 494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000695470222648138,
+      "loss": 0.1136,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006942922340759465,
+      "loss": 0.1108,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006931129738812159,
+      "loss": 0.1083,
+      "step": 497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006919324497820823,
+      "loss": 0.1091,
+      "step": 498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006907506695049534,
+      "loss": 0.1083,
+      "step": 499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006895676407844587,
+      "loss": 0.1083,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006883833713633988,
+      "loss": 0.1077,
+      "step": 501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006871978689926942,
+      "loss": 0.1104,
+      "step": 502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006860111414313349,
+      "loss": 0.107,
+      "step": 503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006848231964463301,
+      "loss": 0.109,
+      "step": 504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006836340418126564,
+      "loss": 0.1093,
+      "step": 505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006824436853132079,
+      "loss": 0.1067,
+      "step": 506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006812521347387443,
+      "loss": 0.1068,
+      "step": 507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006800593978878406,
+      "loss": 0.1067,
+      "step": 508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006788654825668364,
+      "loss": 0.1064,
+      "step": 509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000677670396589783,
+      "loss": 0.1087,
+      "step": 510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006764741477783948,
+      "loss": 0.1073,
+      "step": 511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006752767439619961,
+      "loss": 0.1068,
+      "step": 512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000674078192977471,
+      "loss": 0.1062,
+      "step": 513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006728785026692113,
+      "loss": 0.1066,
+      "step": 514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006716776808890659,
+      "loss": 0.107,
+      "step": 515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006704757354962888,
+      "loss": 0.1058,
+      "step": 516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006692726743574882,
+      "loss": 0.1066,
+      "step": 517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0006680685053465742,
+      "loss": 0.1083,
+      "step": 518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006668632363447087,
+      "loss": 0.1052,
+      "step": 519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006656568752402521,
+      "loss": 0.1066,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000664449429928713,
+      "loss": 0.108,
+      "step": 521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006632409083126958,
+      "loss": 0.1055,
+      "step": 522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006620313183018491,
+      "loss": 0.1066,
+      "step": 523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006608206678128143,
+      "loss": 0.1067,
+      "step": 524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006596089647691734,
+      "loss": 0.11,
+      "step": 525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006583962171013973,
+      "loss": 0.1084,
+      "step": 526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006571824327467937,
+      "loss": 0.1069,
+      "step": 527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006559676196494555,
+      "loss": 0.1049,
+      "step": 528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006547517857602086,
+      "loss": 0.1047,
+      "step": 529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006535349390365597,
+      "loss": 0.1064,
+      "step": 530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006523170874426445,
+      "loss": 0.1062,
+      "step": 531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006510982389491757,
+      "loss": 0.1071,
+      "step": 532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006498784015333901,
+      "loss": 0.1079,
+      "step": 533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006486575831789974,
+      "loss": 0.1087,
+      "step": 534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006474357918761271,
+      "loss": 0.1058,
+      "step": 535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006462130356212767,
+      "loss": 0.1061,
+      "step": 536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006449893224172592,
+      "loss": 0.1079,
+      "step": 537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006437646602731508,
+      "loss": 0.1076,
+      "step": 538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006425390572042381,
+      "loss": 0.1052,
+      "step": 539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006413125212319663,
+      "loss": 0.105,
+      "step": 540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006400850603838865,
+      "loss": 0.1061,
+      "step": 541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006388566826936024,
+      "loss": 0.1055,
+      "step": 542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006376273962007188,
+      "loss": 0.1053,
+      "step": 543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0006363972089507886,
+      "loss": 0.1097,
+      "step": 544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006351661289952595,
+      "loss": 0.1045,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006339341643914225,
+      "loss": 0.1058,
+      "step": 546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006327013232023582,
+      "loss": 0.1081,
+      "step": 547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006314676134968844,
+      "loss": 0.1062,
+      "step": 548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006302330433495036,
+      "loss": 0.1086,
+      "step": 549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006289976208403489,
+      "loss": 0.1063,
+      "step": 550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006277613540551331,
+      "loss": 0.1065,
+      "step": 551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006265242510850941,
+      "loss": 0.1051,
+      "step": 552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006252863200269424,
+      "loss": 0.1055,
+      "step": 553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006240475689828086,
+      "loss": 0.1056,
+      "step": 554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006228080060601904,
+      "loss": 0.1046,
+      "step": 555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006215676393718979,
+      "loss": 0.1051,
+      "step": 556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006203264770360031,
+      "loss": 0.106,
+      "step": 557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006190845271757846,
+      "loss": 0.108,
+      "step": 558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006178417979196757,
+      "loss": 0.106,
+      "step": 559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006165982974012104,
+      "loss": 0.1048,
+      "step": 560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006153540337589709,
+      "loss": 0.1078,
+      "step": 561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000614109015136534,
+      "loss": 0.1032,
+      "step": 562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006128632496824173,
+      "loss": 0.1064,
+      "step": 563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006116167455500265,
+      "loss": 0.1064,
+      "step": 564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006103695108976021,
+      "loss": 0.1063,
+      "step": 565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006091215538881658,
+      "loss": 0.1057,
+      "step": 566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006078728826894668,
+      "loss": 0.1064,
+      "step": 567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006066235054739288,
+      "loss": 0.1042,
+      "step": 568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0006053734304185961,
+      "loss": 0.1051,
+      "step": 569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006041226657050804,
+      "loss": 0.1086,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006028712195195071,
+      "loss": 0.1084,
+      "step": 571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006016191000524619,
+      "loss": 0.1067,
+      "step": 572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0006003663154989369,
+      "loss": 0.1055,
+      "step": 573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005991128740582773,
+      "loss": 0.1074,
+      "step": 574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005978587839341274,
+      "loss": 0.108,
+      "step": 575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005966040533343772,
+      "loss": 0.1064,
+      "step": 576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000595348690471108,
+      "loss": 0.1047,
+      "step": 577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005940927035605403,
+      "loss": 0.1047,
+      "step": 578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005928361008229777,
+      "loss": 0.1054,
+      "step": 579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005915788904827553,
+      "loss": 0.107,
+      "step": 580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005903210807681842,
+      "loss": 0.1056,
+      "step": 581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005890626799114991,
+      "loss": 0.1046,
+      "step": 582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005878036961488028,
+      "loss": 0.1068,
+      "step": 583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005865441377200137,
+      "loss": 0.1047,
+      "step": 584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005852840128688112,
+      "loss": 0.1069,
+      "step": 585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005840233298425818,
+      "loss": 0.104,
+      "step": 586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005827620968923652,
+      "loss": 0.1048,
+      "step": 587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005815003222728007,
+      "loss": 0.1068,
+      "step": 588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005802380142420722,
+      "loss": 0.1059,
+      "step": 589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005789751810618551,
+      "loss": 0.1056,
+      "step": 590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005777118309972614,
+      "loss": 0.1068,
+      "step": 591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005764479723167867,
+      "loss": 0.1067,
+      "step": 592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005751836132922549,
+      "loss": 0.1048,
+      "step": 593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0005739187621987648,
+      "loss": 0.1027,
+      "step": 594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005726534273146358,
+      "loss": 0.1054,
+      "step": 595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005713876169213538,
+      "loss": 0.107,
+      "step": 596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005701213393035163,
+      "loss": 0.105,
+      "step": 597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005688546027487792,
+      "loss": 0.1048,
+      "step": 598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000567587415547802,
+      "loss": 0.1051,
+      "step": 599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005663197859941938,
+      "loss": 0.108,
+      "step": 600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005650517223844585,
+      "loss": 0.1053,
+      "step": 601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005637832330179409,
+      "loss": 0.1033,
+      "step": 602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005625143261967728,
+      "loss": 0.1058,
+      "step": 603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005612450102258175,
+      "loss": 0.105,
+      "step": 604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005599752934126168,
+      "loss": 0.1063,
+      "step": 605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005587051840673355,
+      "loss": 0.1067,
+      "step": 606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005574346905027075,
+      "loss": 0.1033,
+      "step": 607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000556163821033982,
+      "loss": 0.1049,
+      "step": 608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005548925839788675,
+      "loss": 0.1019,
+      "step": 609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005536209876574792,
+      "loss": 0.1094,
+      "step": 610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000552349040392283,
+      "loss": 0.1057,
+      "step": 611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005510767505080418,
+      "loss": 0.1058,
+      "step": 612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005498041263317613,
+      "loss": 0.1036,
+      "step": 613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005485311761926349,
+      "loss": 0.1037,
+      "step": 614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005472579084219893,
+      "loss": 0.1047,
+      "step": 615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005459843313532301,
+      "loss": 0.1065,
+      "step": 616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005447104533217872,
+      "loss": 0.1075,
+      "step": 617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005434362826650603,
+      "loss": 0.103,
+      "step": 618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0005421618277223644,
+      "loss": 0.1061,
+      "step": 619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000540887096834875,
+      "loss": 0.1074,
+      "step": 620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005396120983455735,
+      "loss": 0.1081,
+      "step": 621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005383368405991932,
+      "loss": 0.1063,
+      "step": 622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005370613319421638,
+      "loss": 0.1051,
+      "step": 623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005357855807225573,
+      "loss": 0.105,
+      "step": 624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005345095952900329,
+      "loss": 0.105,
+      "step": 625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005332333839957834,
+      "loss": 0.1054,
+      "step": 626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005319569551924796,
+      "loss": 0.1061,
+      "step": 627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005306803172342154,
+      "loss": 0.1053,
+      "step": 628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005294034784764542,
+      "loss": 0.1053,
+      "step": 629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005281264472759732,
+      "loss": 0.1056,
+      "step": 630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005268492319908092,
+      "loss": 0.1035,
+      "step": 631
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005255718409802041,
+      "loss": 0.1034,
+      "step": 632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005242942826045496,
+      "loss": 0.1027,
+      "step": 633
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005230165652253329,
+      "loss": 0.1028,
+      "step": 634
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005217386972050819,
+      "loss": 0.1023,
+      "step": 635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005204606869073103,
+      "loss": 0.1053,
+      "step": 636
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000519182542696463,
+      "loss": 0.1027,
+      "step": 637
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005179042729378616,
+      "loss": 0.1029,
+      "step": 638
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000516625885997649,
+      "loss": 0.1041,
+      "step": 639
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005153473902427355,
+      "loss": 0.1045,
+      "step": 640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005140687940407431,
+      "loss": 0.1036,
+      "step": 641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005127901057599516,
+      "loss": 0.1029,
+      "step": 642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005115113337692434,
+      "loss": 0.1042,
+      "step": 643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0005102324864380484,
+      "loss": 0.1044,
+      "step": 644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00050895357213629,
+      "loss": 0.1029,
+      "step": 645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005076745992343296,
+      "loss": 0.1055,
+      "step": 646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005063955761029127,
+      "loss": 0.1062,
+      "step": 647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005051165111131129,
+      "loss": 0.104,
+      "step": 648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000503837412636278,
+      "loss": 0.1049,
+      "step": 649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005025582890439751,
+      "loss": 0.1069,
+      "step": 650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005012791487079355,
+      "loss": 0.1052,
+      "step": 651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0005,
+      "loss": 0.1064,
+      "step": 652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004987208512920646,
+      "loss": 0.1032,
+      "step": 653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000497441710956025,
+      "loss": 0.1043,
+      "step": 654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000496162587363722,
+      "loss": 0.1045,
+      "step": 655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004948834888868872,
+      "loss": 0.1035,
+      "step": 656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004936044238970874,
+      "loss": 0.1041,
+      "step": 657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004923254007656704,
+      "loss": 0.1046,
+      "step": 658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00049104642786371,
+      "loss": 0.1055,
+      "step": 659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004897675135619516,
+      "loss": 0.1049,
+      "step": 660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004884886662307566,
+      "loss": 0.1056,
+      "step": 661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004872098942400484,
+      "loss": 0.1058,
+      "step": 662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00048593120595925686,
+      "loss": 0.1042,
+      "step": 663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00048465260975726456,
+      "loss": 0.1039,
+      "step": 664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00048337411400235097,
+      "loss": 0.1027,
+      "step": 665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004820957270621385,
+      "loss": 0.1021,
+      "step": 666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000480817457303537,
+      "loss": 0.1021,
+      "step": 667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004795393130926898,
+      "loss": 0.1045,
+      "step": 668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004782613027949183,
+      "loss": 0.1041,
+      "step": 669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0004769834347746672,
+      "loss": 0.1048,
+      "step": 670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00047570571739545054,
+      "loss": 0.1022,
+      "step": 671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000474428159019796,
+      "loss": 0.1069,
+      "step": 672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00047315076800919095,
+      "loss": 0.1043,
+      "step": 673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000471873552724027,
+      "loss": 0.103,
+      "step": 674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004705965215235459,
+      "loss": 0.1043,
+      "step": 675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004693196827657847,
+      "loss": 0.1055,
+      "step": 676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004680430448075205,
+      "loss": 0.1062,
+      "step": 677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00046676661600421657,
+      "loss": 0.1041,
+      "step": 678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00046549040470996713,
+      "loss": 0.102,
+      "step": 679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004642144192774429,
+      "loss": 0.1039,
+      "step": 680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00046293866805783624,
+      "loss": 0.1026,
+      "step": 681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004616631594008069,
+      "loss": 0.1034,
+      "step": 682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00046038790165442656,
+      "loss": 0.1012,
+      "step": 683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00045911290316512517,
+      "loss": 0.1027,
+      "step": 684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004578381722776357,
+      "loss": 0.1025,
+      "step": 685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00045656371733493977,
+      "loss": 0.1028,
+      "step": 686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004552895466782129,
+      "loss": 0.1041,
+      "step": 687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004540156686467699,
+      "loss": 0.1036,
+      "step": 688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00045274209157801076,
+      "loss": 0.1009,
+      "step": 689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004514688238073651,
+      "loss": 0.1044,
+      "step": 690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004501958736682387,
+      "loss": 0.105,
+      "step": 691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00044892324949195823,
+      "loss": 0.1016,
+      "step": 692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004476509596077172,
+      "loss": 0.1061,
+      "step": 693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004463790123425209,
+      "loss": 0.1047,
+      "step": 694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0004451074160211325,
+      "loss": 0.103,
+      "step": 695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00044383617896601814,
+      "loss": 0.106,
+      "step": 696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00044256530949729245,
+      "loss": 0.1011,
+      "step": 697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00044129481593266465,
+      "loss": 0.1039,
+      "step": 698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004400247065873833,
+      "loss": 0.0998,
+      "step": 699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004387549897741825,
+      "loss": 0.1033,
+      "step": 700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004374856738032273,
+      "loss": 0.1043,
+      "step": 701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004362167669820592,
+      "loss": 0.1026,
+      "step": 702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00043494827761554166,
+      "loss": 0.1053,
+      "step": 703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00043368021400580633,
+      "loss": 0.1013,
+      "step": 704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000432412584452198,
+      "loss": 0.1044,
+      "step": 705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00043114539725122087,
+      "loss": 0.1012,
+      "step": 706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004298786606964838,
+      "loss": 0.1054,
+      "step": 707
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00042861238307864626,
+      "loss": 0.1042,
+      "step": 708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00042734657268536424,
+      "loss": 0.1023,
+      "step": 709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004260812378012353,
+      "loss": 0.1023,
+      "step": 710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004248163867077452,
+      "loss": 0.1019,
+      "step": 711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00042355202768321336,
+      "loss": 0.1039,
+      "step": 712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004222881690027386,
+      "loss": 0.1045,
+      "step": 713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000421024818938145,
+      "loss": 0.1049,
+      "step": 714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041976198575792786,
+      "loss": 0.1037,
+      "step": 715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004184996777271993,
+      "loss": 0.1001,
+      "step": 716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004172379031076348,
+      "loss": 0.1028,
+      "step": 717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041597667015741836,
+      "loss": 0.1047,
+      "step": 718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000414715987131189,
+      "loss": 0.1033,
+      "step": 719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00041345586227998636,
+      "loss": 0.1047,
+      "step": 720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0004121963038511973,
+      "loss": 0.102,
+      "step": 721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00041093732008850106,
+      "loss": 0.1026,
+      "step": 722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004096789192318158,
+      "loss": 0.1048,
+      "step": 723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040842110951724484,
+      "loss": 0.1048,
+      "step": 724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004071638991770224,
+      "loss": 0.1045,
+      "step": 725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004059072964394599,
+      "loss": 0.1023,
+      "step": 726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040465130952889195,
+      "loss": 0.1036,
+      "step": 727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004033959466656229,
+      "loss": 0.1063,
+      "step": 728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0004021412160658726,
+      "loss": 0.103,
+      "step": 729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00040088712594172265,
+      "loss": 0.1048,
+      "step": 730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000399633684501063,
+      "loss": 0.1033,
+      "step": 731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003983808999475381,
+      "loss": 0.1017,
+      "step": 732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00039712878048049285,
+      "loss": 0.104,
+      "step": 733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003958773342949196,
+      "loss": 0.1042,
+      "step": 734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000394626569581404,
+      "loss": 0.1051,
+      "step": 735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003933764945260712,
+      "loss": 0.1029,
+      "step": 736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003921271173105331,
+      "loss": 0.1046,
+      "step": 737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003908784461118341,
+      "loss": 0.1049,
+      "step": 738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003896304891023978,
+      "loss": 0.103,
+      "step": 739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003883832544499735,
+      "loss": 0.1017,
+      "step": 740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00038713675031758275,
+      "loss": 0.1032,
+      "step": 741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00038589098486346607,
+      "loss": 0.1023,
+      "step": 742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000384645966241029,
+      "loss": 0.1027,
+      "step": 743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003834017025987896,
+      "loss": 0.1047,
+      "step": 744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003821582020803244,
+      "loss": 0.1038,
+      "step": 745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0003809154728242154,
+      "loss": 0.1026,
+      "step": 746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00037967352296399695,
+      "loss": 0.1028,
+      "step": 747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00037843236062810214,
+      "loss": 0.1023,
+      "step": 748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003771919939398097,
+      "loss": 0.1019,
+      "step": 749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003759524310171912,
+      "loss": 0.104,
+      "step": 750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003747136799730576,
+      "loss": 0.1008,
+      "step": 751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00037347574891490596,
+      "loss": 0.1043,
+      "step": 752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00037223864594486676,
+      "loss": 0.1021,
+      "step": 753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00037100237915965094,
+      "loss": 0.1029,
+      "step": 754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00036976695665049655,
+      "loss": 0.1013,
+      "step": 755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003685323865031155,
+      "loss": 0.1029,
+      "step": 756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003672986767976418,
+      "loss": 0.103,
+      "step": 757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00036606583560857756,
+      "loss": 0.1025,
+      "step": 758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003648338710047405,
+      "loss": 0.1047,
+      "step": 759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00036360279104921147,
+      "loss": 0.1012,
+      "step": 760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003623726037992811,
+      "loss": 0.1026,
+      "step": 761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003611433173063976,
+      "loss": 0.1042,
+      "step": 762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003599149396161135,
+      "loss": 0.1023,
+      "step": 763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003586874787680335,
+      "loss": 0.1033,
+      "step": 764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003574609427957618,
+      "loss": 0.1026,
+      "step": 765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003562353397268492,
+      "loss": 0.1058,
+      "step": 766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00035501067758274073,
+      "loss": 0.1029,
+      "step": 767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00035378696437872334,
+      "loss": 0.1035,
+      "step": 768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00035256420812387293,
+      "loss": 0.1024,
+      "step": 769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00035134241682100267,
+      "loss": 0.1043,
+      "step": 770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0003501215984666098,
+      "loss": 0.1029,
+      "step": 771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00034890176105082434,
+      "loss": 0.1044,
+      "step": 772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003476829125573554,
+      "loss": 0.1009,
+      "step": 773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00034646506096344025,
+      "loss": 0.1016,
+      "step": 774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003452482142397915,
+      "loss": 0.1047,
+      "step": 775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003440323803505445,
+      "loss": 0.1017,
+      "step": 776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00034281756725320636,
+      "loss": 0.1018,
+      "step": 777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003416037828986027,
+      "loss": 0.1028,
+      "step": 778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003403910352308266,
+      "loss": 0.1025,
+      "step": 779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00033917933218718564,
+      "loss": 0.1054,
+      "step": 780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00033796868169815096,
+      "loss": 0.1031,
+      "step": 781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00033675909168730435,
+      "loss": 0.1051,
+      "step": 782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00033555057007128705,
+      "loss": 0.1057,
+      "step": 783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00033434312475974786,
+      "loss": 0.105,
+      "step": 784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00033313676365529134,
+      "loss": 0.1032,
+      "step": 785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00033193149465342574,
+      "loss": 0.1036,
+      "step": 786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003307273256425118,
+      "loss": 0.1031,
+      "step": 787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003295242645037112,
+      "loss": 0.1013,
+      "step": 788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003283223191109341,
+      "loss": 0.1023,
+      "step": 789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003271214973307887,
+      "loss": 0.1019,
+      "step": 790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032592180702252895,
+      "loss": 0.1022,
+      "step": 791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003247232560380038,
+      "loss": 0.1041,
+      "step": 792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003235258522216052,
+      "loss": 0.1037,
+      "step": 793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00032232960341021697,
+      "loss": 0.1028,
+      "step": 794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0003211345174331638,
+      "loss": 0.1064,
+      "step": 795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031994060211215934,
+      "loss": 0.1051,
+      "step": 796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00031874786526125574,
+      "loss": 0.1036,
+      "step": 797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003175563146867921,
+      "loss": 0.103,
+      "step": 798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003163659581873435,
+      "loss": 0.1021,
+      "step": 799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031517680355366986,
+      "loss": 0.1028,
+      "step": 800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031398885856866513,
+      "loss": 0.1028,
+      "step": 801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003128021310073059,
+      "loss": 0.1043,
+      "step": 802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031161662863660124,
+      "loss": 0.1046,
+      "step": 803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00031043235921554124,
+      "loss": 0.1027,
+      "step": 804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003092493304950467,
+      "loss": 0.1028,
+      "step": 805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00030806755021791797,
+      "loss": 0.1002,
+      "step": 806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003068870261187843,
+      "loss": 0.1023,
+      "step": 807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003057077659240538,
+      "loss": 0.1032,
+      "step": 808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003045297773518621,
+      "loss": 0.1012,
+      "step": 809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00030335306811202245,
+      "loss": 0.1042,
+      "step": 810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00030217764590597496,
+      "loss": 0.1045,
+      "step": 811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0003010035184267359,
+      "loss": 0.1005,
+      "step": 812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029983069335884815,
+      "loss": 0.104,
+      "step": 813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029865917837833023,
+      "loss": 0.1042,
+      "step": 814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002974889811526263,
+      "loss": 0.1046,
+      "step": 815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029632010934055585,
+      "loss": 0.1036,
+      "step": 816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002951525705922641,
+      "loss": 0.1017,
+      "step": 817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002939863725491706,
+      "loss": 0.1016,
+      "step": 818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002928215228439212,
+      "loss": 0.1015,
+      "step": 819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00029165802910033605,
+      "loss": 0.1021,
+      "step": 820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002904958989333618,
+      "loss": 0.1013,
+      "step": 821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0002893351399490194,
+      "loss": 0.1046,
+      "step": 822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002881757597443563,
+      "loss": 0.1016,
+      "step": 823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00028701776590739575,
+      "loss": 0.1027,
+      "step": 824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002858611660170875,
+      "loss": 0.1008,
+      "step": 825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00028470596764325795,
+      "loss": 0.1019,
+      "step": 826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002835521783465608,
+      "loss": 0.102,
+      "step": 827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002823998056784273,
+      "loss": 0.1009,
+      "step": 828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00028124885718101736,
+      "loss": 0.105,
+      "step": 829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00028009934038716937,
+      "loss": 0.1032,
+      "step": 830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002789512628203519,
+      "loss": 0.1016,
+      "step": 831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002778046319946135,
+      "loss": 0.1041,
+      "step": 832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00027665945541453405,
+      "loss": 0.1042,
+      "step": 833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000275515740575176,
+      "loss": 0.1016,
+      "step": 834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002743734949620339,
+      "loss": 0.1045,
+      "step": 835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00027323272605098717,
+      "loss": 0.1027,
+      "step": 836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002720934413082501,
+      "loss": 0.1037,
+      "step": 837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002709556481903234,
+      "loss": 0.1053,
+      "step": 838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002698193541439449,
+      "loss": 0.1019,
+      "step": 839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002686845666060416,
+      "loss": 0.1004,
+      "step": 840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00026755129300368017,
+      "loss": 0.1007,
+      "step": 841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00026641954075401903,
+      "loss": 0.1022,
+      "step": 842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0002652893172642594,
+      "loss": 0.1058,
+      "step": 843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00026416062993159673,
+      "loss": 0.1018,
+      "step": 844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00026303348614317284,
+      "loss": 0.104,
+      "step": 845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00026190789327602696,
+      "loss": 0.1031,
+      "step": 846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00026078385869704793,
+      "loss": 0.1035,
+      "step": 847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00025966138976292523,
+      "loss": 0.1021,
+      "step": 848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002585404938201019,
+      "loss": 0.1024,
+      "step": 849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00025742117820472585,
+      "loss": 0.1019,
+      "step": 850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002563034502426016,
+      "loss": 0.1015,
+      "step": 851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00025518731724914295,
+      "loss": 0.1019,
+      "step": 852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002540727865293247,
+      "loss": 0.0998,
+      "step": 853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00025295986537763487,
+      "loss": 0.1006,
+      "step": 854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00025184856107802713,
+      "loss": 0.103,
+      "step": 855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002507388809038729,
+      "loss": 0.103,
+      "step": 856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00024963083211791394,
+      "loss": 0.1029,
+      "step": 857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002485244219722148,
+      "loss": 0.1023,
+      "step": 858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00024741965770811515,
+      "loss": 0.1013,
+      "step": 859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002463165465561829,
+      "loss": 0.1042,
+      "step": 860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00024521509573616576,
+      "loss": 0.1023,
+      "step": 861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00024411531245694545,
+      "loss": 0.1027,
+      "step": 862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00024301720391648956,
+      "loss": 0.1054,
+      "step": 863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00024192077730180472,
+      "loss": 0.1036,
+      "step": 864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00024082603978888944,
+      "loss": 0.1025,
+      "step": 865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023973299854268743,
+      "loss": 0.1031,
+      "step": 866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023864166071704024,
+      "loss": 0.1018,
+      "step": 867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023755203345464095,
+      "loss": 0.102,
+      "step": 868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023646412388698702,
+      "loss": 0.0998,
+      "step": 869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023537793913433385,
+      "loss": 0.1022,
+      "step": 870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023429348630564812,
+      "loss": 0.1036,
+      "step": 871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0002332107724985612,
+      "loss": 0.1018,
+      "step": 872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00023212980479932278,
+      "loss": 0.1045,
+      "step": 873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00023105059028275465,
+      "loss": 0.1032,
+      "step": 874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00022997313601220355,
+      "loss": 0.0999,
+      "step": 875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00022889744903949622,
+      "loss": 0.1042,
+      "step": 876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00022782353640489245,
+      "loss": 0.1027,
+      "step": 877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00022675140513703923,
+      "loss": 0.1039,
+      "step": 878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00022568106225292445,
+      "loss": 0.1028,
+      "step": 879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00022461251475783156,
+      "loss": 0.1027,
+      "step": 880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00022354576964529304,
+      "loss": 0.1032,
+      "step": 881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00022248083389704516,
+      "loss": 0.0999,
+      "step": 882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00022141771448298204,
+      "loss": 0.1028,
+      "step": 883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00022035641836111013,
+      "loss": 0.1032,
+      "step": 884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00021929695247750248,
+      "loss": 0.1038,
+      "step": 885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00021823932376625362,
+      "loss": 0.1039,
+      "step": 886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00021718353914943394,
+      "loss": 0.1032,
+      "step": 887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00021612960553704408,
+      "loss": 0.103,
+      "step": 888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00021507752982697048,
+      "loss": 0.1061,
+      "step": 889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00021402731890493958,
+      "loss": 0.1036,
+      "step": 890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00021297897964447316,
+      "loss": 0.1024,
+      "step": 891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00021193251890684294,
+      "loss": 0.1021,
+      "step": 892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002108879435410261,
+      "loss": 0.1022,
+      "step": 893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00020984526038366003,
+      "loss": 0.1022,
+      "step": 894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00020880447625899807,
+      "loss": 0.1041,
+      "step": 895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00020776559797886445,
+      "loss": 0.1024,
+      "step": 896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0002067286323426098,
+      "loss": 0.1048,
+      "step": 897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00020569358613706684,
+      "loss": 0.1016,
+      "step": 898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00020466046613650568,
+      "loss": 0.1016,
+      "step": 899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00020362927910258987,
+      "loss": 0.1003,
+      "step": 900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00020260003178433134,
+      "loss": 0.1054,
+      "step": 901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00020157273091804724,
+      "loss": 0.1028,
+      "step": 902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00020054738322731536,
+      "loss": 0.1029,
+      "step": 903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001995239954229301,
+      "loss": 0.1012,
+      "step": 904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001985025742028585,
+      "loss": 0.1011,
+      "step": 905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019748312625219672,
+      "loss": 0.1027,
+      "step": 906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001964656582431259,
+      "loss": 0.1012,
+      "step": 907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019545017683486872,
+      "loss": 0.104,
+      "step": 908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019443668867364579,
+      "loss": 0.1026,
+      "step": 909
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019342520039263207,
+      "loss": 0.1033,
+      "step": 910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019241571861191358,
+      "loss": 0.1012,
+      "step": 911
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019140824993844397,
+      "loss": 0.1026,
+      "step": 912
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019040280096600142,
+      "loss": 0.1049,
+      "step": 913
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018939937827514507,
+      "loss": 0.1041,
+      "step": 914
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018839798843317247,
+      "loss": 0.1025,
+      "step": 915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001873986379940764,
+      "loss": 0.1016,
+      "step": 916
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018640133349850203,
+      "loss": 0.1014,
+      "step": 917
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018540608147370386,
+      "loss": 0.1026,
+      "step": 918
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018441288843350323,
+      "loss": 0.1033,
+      "step": 919
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018342176087824574,
+      "loss": 0.1045,
+      "step": 920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018243270529475847,
+      "loss": 0.1018,
+      "step": 921
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001814457281563078,
+      "loss": 0.1015,
+      "step": 922
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00018046083592255675,
+      "loss": 0.1037,
+      "step": 923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000179478035039523,
+      "loss": 0.103,
+      "step": 924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017849733193953644,
+      "loss": 0.1011,
+      "step": 925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017751873304119742,
+      "loss": 0.1025,
+      "step": 926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000176542244749334,
+      "loss": 0.1049,
+      "step": 927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000175567873454961,
+      "loss": 0.1003,
+      "step": 928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017459562553523767,
+      "loss": 0.104,
+      "step": 929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017362550735342574,
+      "loss": 0.1022,
+      "step": 930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017265752525884832,
+      "loss": 0.1016,
+      "step": 931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00017169168558684784,
+      "loss": 0.1035,
+      "step": 932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001707279946587449,
+      "loss": 0.1049,
+      "step": 933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016976645878179675,
+      "loss": 0.1036,
+      "step": 934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016880708424915603,
+      "loss": 0.1021,
+      "step": 935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016784987733982977,
+      "loss": 0.1037,
+      "step": 936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016689484431863782,
+      "loss": 0.0999,
+      "step": 937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016594199143617251,
+      "loss": 0.1022,
+      "step": 938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016499132492875712,
+      "loss": 0.1018,
+      "step": 939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016404285101840566,
+      "loss": 0.102,
+      "step": 940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016309657591278114,
+      "loss": 0.1012,
+      "step": 941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001621525058051564,
+      "loss": 0.1023,
+      "step": 942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016121064687437236,
+      "loss": 0.1029,
+      "step": 943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00016027100528479815,
+      "loss": 0.1,
+      "step": 944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015933358718629059,
+      "loss": 0.1019,
+      "step": 945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000158398398714154,
+      "loss": 0.1025,
+      "step": 946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015746544598910006,
+      "loss": 0.104,
+      "step": 947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015653473511720761,
+      "loss": 0.1018,
+      "step": 948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00015560627218988294,
+      "loss": 0.102,
+      "step": 949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015468006328381968,
+      "loss": 0.1007,
+      "step": 950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015375611446095906,
+      "loss": 0.105,
+      "step": 951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015283443176845052,
+      "loss": 0.1002,
+      "step": 952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015191502123861183,
+      "loss": 0.1009,
+      "step": 953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00015099788888888938,
+      "loss": 0.1032,
+      "step": 954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001500830407218195,
+      "loss": 0.1026,
+      "step": 955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014917048272498863,
+      "loss": 0.1002,
+      "step": 956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014826022087099428,
+      "loss": 0.1012,
+      "step": 957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014735226111740603,
+      "loss": 0.1004,
+      "step": 958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014644660940672628,
+      "loss": 0.1014,
+      "step": 959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014554327166635173,
+      "loss": 0.1023,
+      "step": 960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014464225380853425,
+      "loss": 0.1025,
+      "step": 961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001437435617303423,
+      "loss": 0.102,
+      "step": 962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001428472013136225,
+      "loss": 0.1012,
+      "step": 963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014195317842496082,
+      "loss": 0.1023,
+      "step": 964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014106149891564447,
+      "loss": 0.1005,
+      "step": 965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014017216862162357,
+      "loss": 0.1005,
+      "step": 966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013928519336347246,
+      "loss": 0.1004,
+      "step": 967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001384005789463524,
+      "loss": 0.1019,
+      "step": 968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013751833115997308,
+      "loss": 0.1019,
+      "step": 969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001366384557785549,
+      "loss": 0.102,
+      "step": 970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000135760958560791,
+      "loss": 0.101,
+      "step": 971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001348858452498098,
+      "loss": 0.1016,
+      "step": 972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013401312157313716,
+      "loss": 0.1009,
+      "step": 973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013314279324265921,
+      "loss": 0.1016,
+      "step": 974
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001322748659545846,
+      "loss": 0.1032,
+      "step": 975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013140934538940753,
+      "loss": 0.1048,
+      "step": 976
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013054623721187042,
+      "loss": 0.1035,
+      "step": 977
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012968554707092682,
+      "loss": 0.1011,
+      "step": 978
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012882728059970466,
+      "loss": 0.1029,
+      "step": 979
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012797144341546884,
+      "loss": 0.1035,
+      "step": 980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012711804111958507,
+      "loss": 0.1036,
+      "step": 981
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001262670792974831,
+      "loss": 0.1024,
+      "step": 982
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012541856351861973,
+      "loss": 0.1033,
+      "step": 983
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012457249933644287,
+      "loss": 0.1019,
+      "step": 984
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012372889228835494,
+      "loss": 0.1011,
+      "step": 985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001228877478956766,
+      "loss": 0.1003,
+      "step": 986
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012204907166361079,
+      "loss": 0.1009,
+      "step": 987
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012121286908120649,
+      "loss": 0.1025,
+      "step": 988
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00012037914562132302,
+      "loss": 0.101,
+      "step": 989
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011954790674059401,
+      "loss": 0.1022,
+      "step": 990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011871915787939174,
+      "loss": 0.1042,
+      "step": 991
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011789290446179168,
+      "loss": 0.1007,
+      "step": 992
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011706915189553696,
+      "loss": 0.1017,
+      "step": 993
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011624790557200254,
+      "loss": 0.102,
+      "step": 994
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001154291708661605,
+      "loss": 0.1031,
+      "step": 995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011461295313654485,
+      "loss": 0.0999,
+      "step": 996
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011379925772521588,
+      "loss": 0.1018,
+      "step": 997
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011298808995772581,
+      "loss": 0.1029,
+      "step": 998
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011217945514308365,
+      "loss": 0.1041,
+      "step": 999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011137335857372045,
+      "loss": 0.102,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011056980552545465,
+      "loss": 0.1034,
+      "step": 1001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001097688012574578,
+      "loss": 0.1049,
+      "step": 1002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00010897035101221975,
+      "loss": 0.1025,
+      "step": 1003
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00010817446001551468,
+      "loss": 0.1047,
+      "step": 1004
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00010738113347636663,
+      "loss": 0.1021,
+      "step": 1005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00010659037658701576,
+      "loss": 0.1008,
+      "step": 1006
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001058021945228837,
+      "loss": 0.1015,
+      "step": 1007
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00010501659244254053,
+      "loss": 0.1034,
+      "step": 1008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00010423357548767043,
+      "loss": 0.1021,
+      "step": 1009
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00010345314878303824,
+      "loss": 0.1012,
+      "step": 1010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00010267531743645587,
+      "loss": 0.1007,
+      "step": 1011
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00010190008653874889,
+      "loss": 0.101,
+      "step": 1012
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00010112746116372324,
+      "loss": 0.1023,
+      "step": 1013
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00010035744636813187,
+      "loss": 0.102,
+      "step": 1014
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.959004719164183e-05,
+      "loss": 0.1025,
+      "step": 1015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.882526865680125e-05,
+      "loss": 0.1022,
+      "step": 1016
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.806311576900633e-05,
+      "loss": 0.1014,
+      "step": 1017
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.730359351646884e-05,
+      "loss": 0.1017,
+      "step": 1018
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.654670687018324e-05,
+      "loss": 0.1036,
+      "step": 1019
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.579246078389403e-05,
+      "loss": 0.1002,
+      "step": 1020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.504086019406372e-05,
+      "loss": 0.1036,
+      "step": 1021
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.429191001984039e-05,
+      "loss": 0.1032,
+      "step": 1022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.354561516302528e-05,
+      "loss": 0.1026,
+      "step": 1023
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.28019805080409e-05,
+      "loss": 0.101,
+      "step": 1024
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.2061010921899e-05,
+      "loss": 0.1003,
+      "step": 1025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.132271125416874e-05,
+      "loss": 0.102,
+      "step": 1026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.05870863369449e-05,
+      "loss": 0.1003,
+      "step": 1027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.985414098481643e-05,
+      "loss": 0.0994,
+      "step": 1028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.912387999483462e-05,
+      "loss": 0.1023,
+      "step": 1029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.839630814648203e-05,
+      "loss": 0.102,
+      "step": 1030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.767143020164103e-05,
+      "loss": 0.103,
+      "step": 1031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.694925090456268e-05,
+      "loss": 0.1029,
+      "step": 1032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.622977498183555e-05,
+      "loss": 0.1023,
+      "step": 1033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.551300714235494e-05,
+      "loss": 0.1025,
+      "step": 1034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.479895207729227e-05,
+      "loss": 0.1007,
+      "step": 1035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.40876144600638e-05,
+      "loss": 0.1005,
+      "step": 1036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.337899894630064e-05,
+      "loss": 0.1033,
+      "step": 1037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.26731101738178e-05,
+      "loss": 0.1012,
+      "step": 1038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.196995276258423e-05,
+      "loss": 0.1018,
+      "step": 1039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.126953131469229e-05,
+      "loss": 0.101,
+      "step": 1040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.057185041432774e-05,
+      "loss": 0.0997,
+      "step": 1041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.987691462773983e-05,
+      "loss": 0.1023,
+      "step": 1042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.918472850321124e-05,
+      "loss": 0.1022,
+      "step": 1043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.849529657102849e-05,
+      "loss": 0.1023,
+      "step": 1044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.780862334345218e-05,
+      "loss": 0.1043,
+      "step": 1045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.712471331468717e-05,
+      "loss": 0.104,
+      "step": 1046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.644357096085397e-05,
+      "loss": 0.1002,
+      "step": 1047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.576520073995857e-05,
+      "loss": 0.1002,
+      "step": 1048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.508960709186385e-05,
+      "loss": 0.1021,
+      "step": 1049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.441679443826021e-05,
+      "loss": 0.1012,
+      "step": 1050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.374676718263679e-05,
+      "loss": 0.101,
+      "step": 1051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.307952971025245e-05,
+      "loss": 0.102,
+      "step": 1052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.241508638810739e-05,
+      "loss": 0.1005,
+      "step": 1053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.175344156491431e-05,
+      "loss": 0.1027,
+      "step": 1054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.109459957106995e-05,
+      "loss": 0.0998,
+      "step": 1055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.043856471862691e-05,
+      "loss": 0.1006,
+      "step": 1056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.97853413012653e-05,
+      "loss": 0.1016,
+      "step": 1057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.913493359426475e-05,
+      "loss": 0.1017,
+      "step": 1058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.848734585447642e-05,
+      "loss": 0.1014,
+      "step": 1059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.784258232029472e-05,
+      "loss": 0.1023,
+      "step": 1060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.720064721163022e-05,
+      "loss": 0.1012,
+      "step": 1061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.656154472988174e-05,
+      "loss": 0.1027,
+      "step": 1062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.592527905790885e-05,
+      "loss": 0.1027,
+      "step": 1063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.529185436000434e-05,
+      "loss": 0.1021,
+      "step": 1064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.466127478186729e-05,
+      "loss": 0.1022,
+      "step": 1065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.403354445057569e-05,
+      "loss": 0.1032,
+      "step": 1066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.340866747455948e-05,
+      "loss": 0.101,
+      "step": 1067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.278664794357369e-05,
+      "loss": 0.1007,
+      "step": 1068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.216748992867177e-05,
+      "loss": 0.1037,
+      "step": 1069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.155119748217874e-05,
+      "loss": 0.1023,
+      "step": 1070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.0937774637664814e-05,
+      "loss": 0.1043,
+      "step": 1071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.0327225409918975e-05,
+      "loss": 0.1026,
+      "step": 1072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.971955379492255e-05,
+      "loss": 0.101,
+      "step": 1073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.911476376982333e-05,
+      "loss": 0.0999,
+      "step": 1074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8512859292909334e-05,
+      "loss": 0.1016,
+      "step": 1075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7913844303583095e-05,
+      "loss": 0.1009,
+      "step": 1076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.7317722722335544e-05,
+      "loss": 0.1014,
+      "step": 1077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.6724498450720695e-05,
+      "loss": 0.1011,
+      "step": 1078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.613417537132992e-05,
+      "loss": 0.103,
+      "step": 1079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.554675734776665e-05,
+      "loss": 0.1024,
+      "step": 1080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.4962248224620925e-05,
+      "loss": 0.0998,
+      "step": 1081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.4380651827444394e-05,
+      "loss": 0.1009,
+      "step": 1082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.380197196272529e-05,
+      "loss": 0.1012,
+      "step": 1083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.322621241786324e-05,
+      "loss": 0.0988,
+      "step": 1084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.265337696114497e-05,
+      "loss": 0.103,
+      "step": 1085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.208346934171898e-05,
+      "loss": 0.1016,
+      "step": 1086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.15164932895717e-05,
+      "loss": 0.1016,
+      "step": 1087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.095245251550257e-05,
+      "loss": 0.1056,
+      "step": 1088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.039135071110007e-05,
+      "loss": 0.1054,
+      "step": 1089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.983319154871741e-05,
+      "loss": 0.1002,
+      "step": 1090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9277978681448345e-05,
+      "loss": 0.0999,
+      "step": 1091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.872571574310369e-05,
+      "loss": 0.1019,
+      "step": 1092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.817640634818704e-05,
+      "loss": 0.1026,
+      "step": 1093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.763005409187154e-05,
+      "loss": 0.1025,
+      "step": 1094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.708666254997601e-05,
+      "loss": 0.1004,
+      "step": 1095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.6546235278941916e-05,
+      "loss": 0.1023,
+      "step": 1096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.600877581580959e-05,
+      "loss": 0.1034,
+      "step": 1097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.547428767819578e-05,
+      "loss": 0.1025,
+      "step": 1098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.494277436426963e-05,
+      "loss": 0.1015,
+      "step": 1099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.4414239352730864e-05,
+      "loss": 0.1028,
+      "step": 1100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.3888686102786344e-05,
+      "loss": 0.1012,
+      "step": 1101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.3366118054127656e-05,
+      "loss": 0.101,
+      "step": 1102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2846538626908574e-05,
+      "loss": 0.101,
+      "step": 1103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.232995122172245e-05,
+      "loss": 0.1002,
+      "step": 1104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.1816359219580494e-05,
+      "loss": 0.1003,
+      "step": 1105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.130576598188907e-05,
+      "loss": 0.1011,
+      "step": 1106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.079817485042797e-05,
+      "loss": 0.1028,
+      "step": 1107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.0293589147328615e-05,
+      "loss": 0.1007,
+      "step": 1108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.979201217505207e-05,
+      "loss": 0.1005,
+      "step": 1109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.929344721636774e-05,
+      "loss": 0.098,
+      "step": 1110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.879789753433155e-05,
+      "loss": 0.1021,
+      "step": 1111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.830536637226495e-05,
+      "loss": 0.1026,
+      "step": 1112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.781585695373313e-05,
+      "loss": 0.1013,
+      "step": 1113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.732937248252472e-05,
+      "loss": 0.1043,
+      "step": 1114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.684591614263e-05,
+      "loss": 0.1032,
+      "step": 1115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.6365491098220685e-05,
+      "loss": 0.103,
+      "step": 1116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.588810049362895e-05,
+      "loss": 0.1023,
+      "step": 1117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.541374745332676e-05,
+      "loss": 0.102,
+      "step": 1118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4942435081905696e-05,
+      "loss": 0.1026,
+      "step": 1119
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.447416646405632e-05,
+      "loss": 0.1008,
+      "step": 1120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.4008944664548236e-05,
+      "loss": 0.1014,
+      "step": 1121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.354677272820994e-05,
+      "loss": 0.1022,
+      "step": 1122
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.308765367990879e-05,
+      "loss": 0.1022,
+      "step": 1123
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.263159052453146e-05,
+      "loss": 0.1026,
+      "step": 1124
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.2178586246964015e-05,
+      "loss": 0.1027,
+      "step": 1125
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.172864381207252e-05,
+      "loss": 0.1021,
+      "step": 1126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.128176616468359e-05,
+      "loss": 0.1011,
+      "step": 1127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.0837956229565144e-05,
+      "loss": 0.1034,
+      "step": 1128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.039721691140712e-05,
+      "loss": 0.103,
+      "step": 1129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.995955109480275e-05,
+      "loss": 0.1042,
+      "step": 1130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9524961644229464e-05,
+      "loss": 0.0997,
+      "step": 1131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.909345140403019e-05,
+      "loss": 0.1015,
+      "step": 1132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8665023198394646e-05,
+      "loss": 0.1013,
+      "step": 1133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8239679831341126e-05,
+      "loss": 0.1026,
+      "step": 1134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7817424086697906e-05,
+      "loss": 0.1021,
+      "step": 1135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.7398258728085046e-05,
+      "loss": 0.1018,
+      "step": 1136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6982186498896388e-05,
+      "loss": 0.1022,
+      "step": 1137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.656921012228153e-05,
+      "loss": 0.102,
+      "step": 1138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.6159332301128103e-05,
+      "loss": 0.1002,
+      "step": 1139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.5752555718043912e-05,
+      "loss": 0.0983,
+      "step": 1140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.534888303533972e-05,
+      "loss": 0.1005,
+      "step": 1141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4948316895011204e-05,
+      "loss": 0.1,
+      "step": 1142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.4550859918722245e-05,
+      "loss": 0.1021,
+      "step": 1143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.415651470778768e-05,
+      "loss": 0.1036,
+      "step": 1144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3765283843156014e-05,
+      "loss": 0.1021,
+      "step": 1145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3377169885392734e-05,
+      "loss": 0.1036,
+      "step": 1146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.2992175374663416e-05,
+      "loss": 0.1056,
+      "step": 1147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.26103028307173e-05,
+      "loss": 0.1006,
+      "step": 1148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.2231554752870585e-05,
+      "loss": 0.1014,
+      "step": 1149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.1855933619990166e-05,
+      "loss": 0.1034,
+      "step": 1150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.148344189047746e-05,
+      "loss": 0.1008,
+      "step": 1151
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.1114082002252287e-05,
+      "loss": 0.1022,
+      "step": 1152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0747856372736796e-05,
+      "loss": 0.1014,
+      "step": 1153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.038476739883982e-05,
+      "loss": 0.1021,
+      "step": 1154
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.0024817456941012e-05,
+      "loss": 0.1006,
+      "step": 1155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.96680089028754e-05,
+      "loss": 0.1,
+      "step": 1156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.931434407191812e-05,
+      "loss": 0.1007,
+      "step": 1157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8963825278768776e-05,
+      "loss": 0.102,
+      "step": 1158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8616454817536544e-05,
+      "loss": 0.1016,
+      "step": 1159
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.827223496172509e-05,
+      "loss": 0.1006,
+      "step": 1160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.793116796421773e-05,
+      "loss": 0.1028,
+      "step": 1161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7593256057262642e-05,
+      "loss": 0.103,
+      "step": 1162
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7258501452458163e-05,
+      "loss": 0.1017,
+      "step": 1163
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6926906340738568e-05,
+      "loss": 0.1031,
+      "step": 1164
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6598472892359497e-05,
+      "loss": 0.101,
+      "step": 1165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.627320325688375e-05,
+      "loss": 0.1004,
+      "step": 1166
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5951099563167593e-05,
+      "loss": 0.1011,
+      "step": 1167
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5632163919346076e-05,
+      "loss": 0.1015,
+      "step": 1168
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.531639841282001e-05,
+      "loss": 0.1011,
+      "step": 1169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.5003805110241964e-05,
+      "loss": 0.1015,
+      "step": 1170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.4694386057502728e-05,
+      "loss": 0.1023,
+      "step": 1171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.438814327971788e-05,
+      "loss": 0.1003,
+      "step": 1172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.4085078781214733e-05,
+      "loss": 0.102,
+      "step": 1173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3785194545518964e-05,
+      "loss": 0.1042,
+      "step": 1174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.348849253534179e-05,
+      "loss": 0.1031,
+      "step": 1175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3194974692567252e-05,
+      "loss": 0.1028,
+      "step": 1176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2904642938239119e-05,
+      "loss": 0.1018,
+      "step": 1177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2617499172548619e-05,
+      "loss": 0.1017,
+      "step": 1178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2333545274822001e-05,
+      "loss": 0.1008,
+      "step": 1179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2052783103508103e-05,
+      "loss": 0.1015,
+      "step": 1180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1775214496166253e-05,
+      "loss": 0.1,
+      "step": 1181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1500841269454166e-05,
+      "loss": 0.1,
+      "step": 1182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.122966521911617e-05,
+      "loss": 0.1023,
+      "step": 1183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0961688119971447e-05,
+      "loss": 0.1009,
+      "step": 1184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.069691172590237e-05,
+      "loss": 0.1015,
+      "step": 1185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0435337769843012e-05,
+      "loss": 0.1025,
+      "step": 1186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0176967963767713e-05,
+      "loss": 0.1022,
+      "step": 1187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.9218039986802e-06,
+      "loss": 0.0997,
+      "step": 1188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.669847544602095e-06,
+      "loss": 0.1039,
+      "step": 1189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.421100250562309e-06,
+      "loss": 0.1023,
+      "step": 1190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.175563744586113e-06,
+      "loss": 0.1018,
+      "step": 1191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.933239633684531e-06,
+      "loss": 0.1021,
+      "step": 1192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.694129503843684e-06,
+      "loss": 0.1031,
+      "step": 1193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.458234920014684e-06,
+      "loss": 0.1004,
+      "step": 1194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.225557426103036e-06,
+      "loss": 0.1003,
+      "step": 1195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.996098544958863e-06,
+      "loss": 0.1007,
+      "step": 1196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.76985977836675e-06,
+      "loss": 0.1019,
+      "step": 1197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.5468426070360865e-06,
+      "loss": 0.1024,
+      "step": 1198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.327048490591015e-06,
+      "loss": 0.102,
+      "step": 1199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 7.110478867561332e-06,
+      "loss": 0.1001,
+      "step": 1200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.897135155372769e-06,
+      "loss": 0.1015,
+      "step": 1201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.687018750337726e-06,
+      "loss": 0.1027,
+      "step": 1202
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.480131027646441e-06,
+      "loss": 0.103,
+      "step": 1203
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.276473341357558e-06,
+      "loss": 0.1014,
+      "step": 1204
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.076047024389519e-06,
+      "loss": 0.1003,
+      "step": 1205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8788533885117956e-06,
+      "loss": 0.1018,
+      "step": 1206
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.684893724336337e-06,
+      "loss": 0.1009,
+      "step": 1207
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.494169301309027e-06,
+      "loss": 0.1023,
+      "step": 1208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3066813677014045e-06,
+      "loss": 0.0996,
+      "step": 1209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.122431150602624e-06,
+      "loss": 0.1017,
+      "step": 1210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941419855911233e-06,
+      "loss": 0.1023,
+      "step": 1211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.763648668327458e-06,
+      "loss": 0.0996,
+      "step": 1212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5891187513452114e-06,
+      "loss": 0.1022,
+      "step": 1213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.417831247244819e-06,
+      "loss": 0.1012,
+      "step": 1214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.249787277085138e-06,
+      "loss": 0.0996,
+      "step": 1215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.084987940696561e-06,
+      "loss": 0.0995,
+      "step": 1216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.923434316673746e-06,
+      "loss": 0.101,
+      "step": 1217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.765127462368345e-06,
+      "loss": 0.1024,
+      "step": 1218
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.6100684138823945e-06,
+      "loss": 0.1009,
+      "step": 1219
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.458258186061214e-06,
+      "loss": 0.1012,
+      "step": 1220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.309697772487019e-06,
+      "loss": 0.101,
+      "step": 1221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.1643881454723744e-06,
+      "loss": 0.1013,
+      "step": 1222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.022330256053696e-06,
+      "loss": 0.1023,
+      "step": 1223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.883525033985146e-06,
+      "loss": 0.1004,
+      "step": 1224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.747973387732583e-06,
+      "loss": 0.1025,
+      "step": 1225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.615676204467343e-06,
+      "loss": 0.1005,
+      "step": 1226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.4866343500609677e-06,
+      "loss": 0.1012,
+      "step": 1227
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.3608486690788746e-06,
+      "loss": 0.105,
+      "step": 1228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.238319984775361e-06,
+      "loss": 0.1003,
+      "step": 1229
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.1190490990879994e-06,
+      "loss": 0.1008,
+      "step": 1230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.0030367926323624e-06,
+      "loss": 0.105,
+      "step": 1231
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8902838246969145e-06,
+      "loss": 0.0996,
+      "step": 1232
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7807909332382411e-06,
+      "loss": 0.1026,
+      "step": 1233
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.6745588348758833e-06,
+      "loss": 0.1035,
+      "step": 1234
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5715882248880098e-06,
+      "loss": 0.1037,
+      "step": 1235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.4718797772065306e-06,
+      "loss": 0.1026,
+      "step": 1236
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3754341444128792e-06,
+      "loss": 0.1021,
+      "step": 1237
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2822519577337932e-06,
+      "loss": 0.1,
+      "step": 1238
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1923338270369844e-06,
+      "loss": 0.1016,
+      "step": 1239
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1056803408273086e-06,
+      "loss": 0.0999,
+      "step": 1240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0222920662428248e-06,
+      "loss": 0.1009,
+      "step": 1241
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.421695490512416e-07,
+      "loss": 0.1026,
+      "step": 1242
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.653133136460877e-07,
+      "loss": 0.1012,
+      "step": 1243
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.917238630433809e-07,
+      "loss": 0.103,
+      "step": 1244
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 7.214016788785749e-07,
+      "loss": 0.1032,
+      "step": 1245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.543472214030066e-07,
+      "loss": 0.1026,
+      "step": 1246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.905609294811765e-07,
+      "loss": 0.1025,
+      "step": 1247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.300432205876948e-07,
+      "loss": 0.1017,
+      "step": 1248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7279449080478434e-07,
+      "loss": 0.1002,
+      "step": 1249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1881511481939304e-07,
+      "loss": 0.1027,
+      "step": 1250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.6810544592091876e-07,
+      "loss": 0.1032,
+      "step": 1251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.2066581599898837e-07,
+      "loss": 0.1016,
+      "step": 1252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.764965355410154e-07,
+      "loss": 0.0994,
+      "step": 1253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.355978936303127e-07,
+      "loss": 0.1021,
+      "step": 1254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9797015794426053e-07,
+      "loss": 0.1014,
+      "step": 1255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.6361357475258577e-07,
+      "loss": 0.1004,
+      "step": 1256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3252836891547437e-07,
+      "loss": 0.0998,
+      "step": 1257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0471474388240587e-07,
+      "loss": 0.1,
+      "step": 1258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.017288169070991e-08,
+      "loss": 0.099,
+      "step": 1259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8902942964289554e-08,
+      "loss": 0.099,
+      "step": 1260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.090506691262208e-08,
+      "loss": 0.1009,
+      "step": 1261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.617937133009285e-08,
+      "loss": 0.1011,
+      "step": 1262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.472595259477405e-08,
+      "loss": 0.1029,
+      "step": 1263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 6.5448856680361265e-09,
+      "loss": 0.0996,
+      "step": 1264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.636224094214711e-09,
+      "loss": 0.1011,
+      "step": 1265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0,
+      "loss": 0.1027,
+      "step": 1266
+    },
+    {
+      "epoch": 0.5,
+      "step": 1266,
+      "total_flos": 251756715245568.0,
+      "train_loss": 0.15758075081341638,
+      "train_runtime": 17032.1168,
+      "train_samples_per_second": 9.511,
+      "train_steps_per_second": 0.074
+    }
+  ],
+  "max_steps": 1266,
+  "num_train_epochs": 1,
+  "total_flos": 251756715245568.0,
+  "trial_name": null,
+  "trial_params": null
+}