diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14094 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 2344,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8169014084507043e-07,
+      "loss": 4.5039,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.633802816901409e-07,
+      "loss": 4.5234,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.450704225352114e-07,
+      "loss": 4.5273,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1267605633802817e-06,
+      "loss": 4.5273,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4084507042253523e-06,
+      "loss": 4.4805,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6901408450704227e-06,
+      "loss": 4.3711,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.971830985915493e-06,
+      "loss": 3.7793,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2535211267605635e-06,
+      "loss": 3.6855,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.535211267605634e-06,
+      "loss": 3.0762,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8169014084507046e-06,
+      "loss": 2.0723,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0985915492957746e-06,
+      "loss": 1.834,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3802816901408454e-06,
+      "loss": 1.3809,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6619718309859158e-06,
+      "loss": 0.9639,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.943661971830986e-06,
+      "loss": 0.7783,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.225352112676057e-06,
+      "loss": 0.5115,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.507042253521127e-06,
+      "loss": 0.3538,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.788732394366197e-06,
+      "loss": 0.2145,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.070422535211268e-06,
+      "loss": 0.2125,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.352112676056338e-06,
+      "loss": 0.1752,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.633802816901409e-06,
+      "loss": 0.1571,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.915492957746479e-06,
+      "loss": 0.1566,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.197183098591549e-06,
+      "loss": 0.1469,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.478873239436621e-06,
+      "loss": 0.152,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.760563380281691e-06,
+      "loss": 0.1613,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.042253521126761e-06,
+      "loss": 0.1362,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.3239436619718316e-06,
+      "loss": 0.145,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.6056338028169015e-06,
+      "loss": 0.1523,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.887323943661972e-06,
+      "loss": 0.1499,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.169014084507043e-06,
+      "loss": 0.1378,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.450704225352114e-06,
+      "loss": 0.1379,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.732394366197183e-06,
+      "loss": 0.1433,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.014084507042254e-06,
+      "loss": 0.1357,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.295774647887325e-06,
+      "loss": 0.1365,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.577464788732394e-06,
+      "loss": 0.1329,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.859154929577466e-06,
+      "loss": 0.1395,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0140845070422535e-05,
+      "loss": 0.1493,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0422535211267606e-05,
+      "loss": 0.1361,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0704225352112675e-05,
+      "loss": 0.1288,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0985915492957748e-05,
+      "loss": 0.1268,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1267605633802819e-05,
+      "loss": 0.132,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1549295774647888e-05,
+      "loss": 0.1481,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1830985915492958e-05,
+      "loss": 0.126,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.211267605633803e-05,
+      "loss": 0.1392,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2394366197183098e-05,
+      "loss": 0.1483,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2676056338028171e-05,
+      "loss": 0.1345,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2957746478873242e-05,
+      "loss": 0.1299,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.323943661971831e-05,
+      "loss": 0.1365,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3521126760563382e-05,
+      "loss": 0.141,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.380281690140845e-05,
+      "loss": 0.1246,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4084507042253522e-05,
+      "loss": 0.1293,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4366197183098594e-05,
+      "loss": 0.1348,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4647887323943663e-05,
+      "loss": 0.1429,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4929577464788734e-05,
+      "loss": 0.1118,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5211267605633803e-05,
+      "loss": 0.1243,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5492957746478872e-05,
+      "loss": 0.1324,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5774647887323945e-05,
+      "loss": 0.127,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6056338028169017e-05,
+      "loss": 0.1247,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6338028169014086e-05,
+      "loss": 0.1167,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6619718309859155e-05,
+      "loss": 0.1238,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6901408450704228e-05,
+      "loss": 0.1207,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7183098591549297e-05,
+      "loss": 0.1255,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7464788732394366e-05,
+      "loss": 0.1185,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.774647887323944e-05,
+      "loss": 0.1255,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8028169014084508e-05,
+      "loss": 0.1403,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.830985915492958e-05,
+      "loss": 0.1298,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.859154929577465e-05,
+      "loss": 0.1259,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.887323943661972e-05,
+      "loss": 0.1113,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9154929577464788e-05,
+      "loss": 0.1293,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.943661971830986e-05,
+      "loss": 0.1187,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9718309859154933e-05,
+      "loss": 0.1298,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2e-05,
+      "loss": 0.1216,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999990448516957e-05,
+      "loss": 0.1188,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999961794086065e-05,
+      "loss": 0.1166,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999914036762066e-05,
+      "loss": 0.1157,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9999847176636188e-05,
+      "loss": 0.1138,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999976121383616e-05,
+      "loss": 0.1299,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999965614852619e-05,
+      "loss": 0.1259,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999953198090699e-05,
+      "loss": 0.1079,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999388711215752e-05,
+      "loss": 0.1166,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999226339726166e-05,
+      "loss": 0.1168,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9999044866748405e-05,
+      "loss": 0.133,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998844292629144e-05,
+      "loss": 0.1205,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999862461775153e-05,
+      "loss": 0.1114,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9998385842535216e-05,
+      "loss": 0.1255,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999812796743633e-05,
+      "loss": 0.1116,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.999785099294749e-05,
+      "loss": 0.1221,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99975549195978e-05,
+      "loss": 0.1076,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9997239747952846e-05,
+      "loss": 0.1137,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996905478614698e-05,
+      "loss": 0.1254,
+      "step": 89
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999655211222191e-05,
+      "loss": 0.1121,
+      "step": 90
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9996179649449525e-05,
+      "loss": 0.1115,
+      "step": 91
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995788091009048e-05,
+      "loss": 0.1221,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9995377437648472e-05,
+      "loss": 0.1132,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994947690152273e-05,
+      "loss": 0.117,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994498849341394e-05,
+      "loss": 0.1156,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9994030916073248e-05,
+      "loss": 0.1211,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9993543891241733e-05,
+      "loss": 0.1188,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.999303777577721e-05,
+      "loss": 0.1056,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9992512570646508e-05,
+      "loss": 0.1058,
+      "step": 99
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9991968276852926e-05,
+      "loss": 0.1047,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999140489543622e-05,
+      "loss": 0.1023,
+      "step": 101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990822427472624e-05,
+      "loss": 0.1014,
+      "step": 102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990220874074825e-05,
+      "loss": 0.1,
+      "step": 103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989600236391963e-05,
+      "loss": 0.1149,
+      "step": 104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988960515609645e-05,
+      "loss": 0.0876,
+      "step": 105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998830171294992e-05,
+      "loss": 0.1041,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9987623829671307e-05,
+      "loss": 0.0958,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986926867068754e-05,
+      "loss": 0.092,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986210826473675e-05,
+      "loss": 0.0828,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998547570925391e-05,
+      "loss": 0.0918,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998472151681376e-05,
+      "loss": 0.0797,
+      "step": 111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983948250593954e-05,
+      "loss": 0.1138,
+      "step": 112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9983155912071656e-05,
+      "loss": 0.0928,
+      "step": 113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.998234450276047e-05,
+      "loss": 0.0984,
+      "step": 114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9981514024210428e-05,
+      "loss": 0.0942,
+      "step": 115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9980664478007995e-05,
+      "loss": 0.1046,
+      "step": 116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.997979586577605e-05,
+      "loss": 0.0996,
+      "step": 117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978908189173904e-05,
+      "loss": 0.0938,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9978001449897278e-05,
+      "loss": 0.0947,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9977075649678316e-05,
+      "loss": 0.0956,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9976130790285574e-05,
+      "loss": 0.0775,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9975166873524006e-05,
+      "loss": 0.1052,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9974183901234988e-05,
+      "loss": 0.0767,
+      "step": 123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997318187529628e-05,
+      "loss": 0.0887,
+      "step": 124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.997216079762206e-05,
+      "loss": 0.0811,
+      "step": 125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9971120670162873e-05,
+      "loss": 0.0802,
+      "step": 126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9970061494905685e-05,
+      "loss": 0.0761,
+      "step": 127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996898327387383e-05,
+      "loss": 0.0811,
+      "step": 128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9967886009127028e-05,
+      "loss": 0.0825,
+      "step": 129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9966769702761385e-05,
+      "loss": 0.0861,
+      "step": 130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996563435690937e-05,
+      "loss": 0.082,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9964479973739838e-05,
+      "loss": 0.072,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9963306555458e-05,
+      "loss": 0.0743,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.996211410430543e-05,
+      "loss": 0.0959,
+      "step": 134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.996090262256007e-05,
+      "loss": 0.0729,
+      "step": 135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959672112536202e-05,
+      "loss": 0.0818,
+      "step": 136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9958422576584474e-05,
+      "loss": 0.0916,
+      "step": 137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995715401709186e-05,
+      "loss": 0.0769,
+      "step": 138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9955866436481695e-05,
+      "loss": 0.0737,
+      "step": 139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9954559837213628e-05,
+      "loss": 0.0776,
+      "step": 140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9953234221783665e-05,
+      "loss": 0.0826,
+      "step": 141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9951889592724113e-05,
+      "loss": 0.0781,
+      "step": 142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.995052595260362e-05,
+      "loss": 0.0803,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9949143304027138e-05,
+      "loss": 0.097,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.994774164963594e-05,
+      "loss": 0.0815,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99463209921076e-05,
+      "loss": 0.0852,
+      "step": 146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9944881334155993e-05,
+      "loss": 0.0818,
+      "step": 147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9943422678531297e-05,
+      "loss": 0.0818,
+      "step": 148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9941945028019975e-05,
+      "loss": 0.0842,
+      "step": 149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9940448385444773e-05,
+      "loss": 0.0655,
+      "step": 150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.993893275366473e-05,
+      "loss": 0.0619,
+      "step": 151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9937398135575146e-05,
+      "loss": 0.0759,
+      "step": 152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9935844534107603e-05,
+      "loss": 0.0691,
+      "step": 153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9934271952229937e-05,
+      "loss": 0.0727,
+      "step": 154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9932680392946243e-05,
+      "loss": 0.0559,
+      "step": 155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9931069859296875e-05,
+      "loss": 0.0706,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9929440354358428e-05,
+      "loss": 0.0642,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9927791881243745e-05,
+      "loss": 0.0677,
+      "step": 158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992612444310189e-05,
+      "loss": 0.0718,
+      "step": 159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9924438043118174e-05,
+      "loss": 0.0681,
+      "step": 160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9922732684514114e-05,
+      "loss": 0.0728,
+      "step": 161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.992100837054745e-05,
+      "loss": 0.0588,
+      "step": 162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991926510451214e-05,
+      "loss": 0.0678,
+      "step": 163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9917502889738335e-05,
+      "loss": 0.065,
+      "step": 164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991572172959239e-05,
+      "loss": 0.0861,
+      "step": 165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991392162747684e-05,
+      "loss": 0.0646,
+      "step": 166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9912102586830427e-05,
+      "loss": 0.0703,
+      "step": 167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.991026461112805e-05,
+      "loss": 0.0648,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9908407703880798e-05,
+      "loss": 0.0649,
+      "step": 169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9906531868635897e-05,
+      "loss": 0.0833,
+      "step": 170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9904637108976762e-05,
+      "loss": 0.0605,
+      "step": 171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9902723428522944e-05,
+      "loss": 0.056,
+      "step": 172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9900790830930136e-05,
+      "loss": 0.0719,
+      "step": 173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9898839319890178e-05,
+      "loss": 0.0669,
+      "step": 174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9896868899131032e-05,
+      "loss": 0.0746,
+      "step": 175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9894879572416786e-05,
+      "loss": 0.0587,
+      "step": 176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9892871343547647e-05,
+      "loss": 0.0734,
+      "step": 177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9890844216359923e-05,
+      "loss": 0.0603,
+      "step": 178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988879819472603e-05,
+      "loss": 0.0616,
+      "step": 179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9886733282554478e-05,
+      "loss": 0.0662,
+      "step": 180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9884649483789858e-05,
+      "loss": 0.0551,
+      "step": 181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9882546802412848e-05,
+      "loss": 0.087,
+      "step": 182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.988042524244019e-05,
+      "loss": 0.0691,
+      "step": 183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9878284807924696e-05,
+      "loss": 0.0634,
+      "step": 184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.987612550295523e-05,
+      "loss": 0.0634,
+      "step": 185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.98739473316567e-05,
+      "loss": 0.0709,
+      "step": 186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9871750298190067e-05,
+      "loss": 0.0633,
+      "step": 187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9869534406752312e-05,
+      "loss": 0.0745,
+      "step": 188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986729966157644e-05,
+      "loss": 0.0505,
+      "step": 189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.986504606693149e-05,
+      "loss": 0.0682,
+      "step": 190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9862773627122488e-05,
+      "loss": 0.0626,
+      "step": 191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9860482346490467e-05,
+      "loss": 0.0604,
+      "step": 192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9858172229412452e-05,
+      "loss": 0.0595,
+      "step": 193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.985584328030146e-05,
+      "loss": 0.0574,
+      "step": 194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9853495503606466e-05,
+      "loss": 0.0816,
+      "step": 195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9851128903812422e-05,
+      "loss": 0.0586,
+      "step": 196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984874348544024e-05,
+      "loss": 0.0497,
+      "step": 197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984633925304677e-05,
+      "loss": 0.0724,
+      "step": 198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984391621122481e-05,
+      "loss": 0.0594,
+      "step": 199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9841474364603095e-05,
+      "loss": 0.0686,
+      "step": 200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.983901371784627e-05,
+      "loss": 0.0633,
+      "step": 201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.98365342756549e-05,
+      "loss": 0.0654,
+      "step": 202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9834036042765462e-05,
+      "loss": 0.0674,
+      "step": 203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9831519023950317e-05,
+      "loss": 0.0782,
+      "step": 204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9828983224017714e-05,
+      "loss": 0.0614,
+      "step": 205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9826428647811788e-05,
+      "loss": 0.051,
+      "step": 206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9823855300212534e-05,
+      "loss": 0.0741,
+      "step": 207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982126318613581e-05,
+      "loss": 0.0605,
+      "step": 208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9818652310533326e-05,
+      "loss": 0.0706,
+      "step": 209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981602267839263e-05,
+      "loss": 0.0668,
+      "step": 210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9813374294737092e-05,
+      "loss": 0.0468,
+      "step": 211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.981070716462591e-05,
+      "loss": 0.0592,
+      "step": 212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9808021293154103e-05,
+      "loss": 0.0603,
+      "step": 213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9805316685452477e-05,
+      "loss": 0.058,
+      "step": 214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9802593346687634e-05,
+      "loss": 0.0514,
+      "step": 215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9799851282061957e-05,
+      "loss": 0.0497,
+      "step": 216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979709049681361e-05,
+      "loss": 0.0692,
+      "step": 217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9794310996216505e-05,
+      "loss": 0.0737,
+      "step": 218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9791512785580312e-05,
+      "loss": 0.0527,
+      "step": 219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978869587025045e-05,
+      "loss": 0.0595,
+      "step": 220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9785860255608055e-05,
+      "loss": 0.0551,
+      "step": 221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.978300594707e-05,
+      "loss": 0.0779,
+      "step": 222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9780132950088855e-05,
+      "loss": 0.0458,
+      "step": 223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9777241270152898e-05,
+      "loss": 0.0533,
+      "step": 224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9774330912786096e-05,
+      "loss": 0.0853,
+      "step": 225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.977140188354809e-05,
+      "loss": 0.0724,
+      "step": 226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9768454188034206e-05,
+      "loss": 0.0583,
+      "step": 227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9765487831875405e-05,
+      "loss": 0.0555,
+      "step": 228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9762502820738315e-05,
+      "loss": 0.0788,
+      "step": 229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9759499160325187e-05,
+      "loss": 0.0701,
+      "step": 230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9756476856373905e-05,
+      "loss": 0.0596,
+      "step": 231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.975343591465797e-05,
+      "loss": 0.0522,
+      "step": 232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9750376340986473e-05,
+      "loss": 0.0722,
+      "step": 233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974729814120412e-05,
+      "loss": 0.0805,
+      "step": 234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9744201321191176e-05,
+      "loss": 0.0558,
+      "step": 235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.974108588686349e-05,
+      "loss": 0.0691,
+      "step": 236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9737951844172465e-05,
+      "loss": 0.0622,
+      "step": 237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.973479919910505e-05,
+      "loss": 0.0576,
+      "step": 238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9731627957683737e-05,
+      "loss": 0.0506,
+      "step": 239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9728438125966535e-05,
+      "loss": 0.0447,
+      "step": 240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9725229710046965e-05,
+      "loss": 0.058,
+      "step": 241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9722002716054055e-05,
+      "loss": 0.0608,
+      "step": 242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9718757150152325e-05,
+      "loss": 0.0608,
+      "step": 243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9715493018541763e-05,
+      "loss": 0.077,
+      "step": 244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9712210327457836e-05,
+      "loss": 0.0487,
+      "step": 245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.970890908317145e-05,
+      "loss": 0.0497,
+      "step": 246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9705589291988964e-05,
+      "loss": 0.0607,
+      "step": 247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9702250960252164e-05,
+      "loss": 0.0409,
+      "step": 248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9698894094338253e-05,
+      "loss": 0.069,
+      "step": 249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969551870065984e-05,
+      "loss": 0.0706,
+      "step": 250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.969212478566493e-05,
+      "loss": 0.0576,
+      "step": 251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9688712355836904e-05,
+      "loss": 0.0643,
+      "step": 252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968528141769452e-05,
+      "loss": 0.0658,
+      "step": 253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.968183197779188e-05,
+      "loss": 0.0636,
+      "step": 254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9678364042718446e-05,
+      "loss": 0.07,
+      "step": 255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9674877619099e-05,
+      "loss": 0.0468,
+      "step": 256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9671372713593643e-05,
+      "loss": 0.0513,
+      "step": 257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.966784933289778e-05,
+      "loss": 0.0608,
+      "step": 258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9664307483742125e-05,
+      "loss": 0.0481,
+      "step": 259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9660747172892653e-05,
+      "loss": 0.083,
+      "step": 260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9657168407150612e-05,
+      "loss": 0.0479,
+      "step": 261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.965357119335251e-05,
+      "loss": 0.0663,
+      "step": 262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.964995553837009e-05,
+      "loss": 0.0605,
+      "step": 263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9646321449110326e-05,
+      "loss": 0.048,
+      "step": 264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9642668932515407e-05,
+      "loss": 0.0492,
+      "step": 265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9638997995562724e-05,
+      "loss": 0.0648,
+      "step": 266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963530864526485e-05,
+      "loss": 0.0579,
+      "step": 267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.963160088866955e-05,
+      "loss": 0.0531,
+      "step": 268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9627874732859724e-05,
+      "loss": 0.0647,
+      "step": 269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962413018495345e-05,
+      "loss": 0.0623,
+      "step": 270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.962036725210391e-05,
+      "loss": 0.0716,
+      "step": 271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9616585941499432e-05,
+      "loss": 0.0532,
+      "step": 272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.961278626036344e-05,
+      "loss": 0.0781,
+      "step": 273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9608968215954445e-05,
+      "loss": 0.0443,
+      "step": 274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.960513181556605e-05,
+      "loss": 0.0501,
+      "step": 275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9601277066526922e-05,
+      "loss": 0.0609,
+      "step": 276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9597403976200767e-05,
+      "loss": 0.0702,
+      "step": 277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.959351255198634e-05,
+      "loss": 0.0546,
+      "step": 278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9589602801317415e-05,
+      "loss": 0.0641,
+      "step": 279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9585674731662777e-05,
+      "loss": 0.0599,
+      "step": 280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95817283505262e-05,
+      "loss": 0.048,
+      "step": 281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9577763665446444e-05,
+      "loss": 0.0505,
+      "step": 282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9573780683997236e-05,
+      "loss": 0.0795,
+      "step": 283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.956977941378725e-05,
+      "loss": 0.0711,
+      "step": 284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.95657598624601e-05,
+      "loss": 0.0752,
+      "step": 285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9561722037694318e-05,
+      "loss": 0.0742,
+      "step": 286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.955766594720335e-05,
+      "loss": 0.0459,
+      "step": 287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9553591598735533e-05,
+      "loss": 0.0655,
+      "step": 288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954949900007408e-05,
+      "loss": 0.0569,
+      "step": 289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9545388159037064e-05,
+      "loss": 0.0533,
+      "step": 290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954125908347742e-05,
+      "loss": 0.0748,
+      "step": 291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95371117812829e-05,
+      "loss": 0.0556,
+      "step": 292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9532946260376078e-05,
+      "loss": 0.0682,
+      "step": 293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9528762528714345e-05,
+      "loss": 0.0627,
+      "step": 294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9524560594289863e-05,
+      "loss": 0.0606,
+      "step": 295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952034046512957e-05,
+      "loss": 0.0436,
+      "step": 296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951610214929517e-05,
+      "loss": 0.0611,
+      "step": 297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95118456548831e-05,
+      "loss": 0.0622,
+      "step": 298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9507570990024534e-05,
+      "loss": 0.0647,
+      "step": 299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.950327816288534e-05,
+      "loss": 0.0456,
+      "step": 300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9498967181666094e-05,
+      "loss": 0.0691,
+      "step": 301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9494638054602053e-05,
+      "loss": 0.0677,
+      "step": 302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.949029078996313e-05,
+      "loss": 0.0491,
+      "step": 303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.948592539605389e-05,
+      "loss": 0.0556,
+      "step": 304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9481541881213534e-05,
+      "loss": 0.0596,
+      "step": 305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.947714025381587e-05,
+      "loss": 0.0531,
+      "step": 306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9472720522269316e-05,
+      "loss": 0.0606,
+      "step": 307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9468282695016864e-05,
+      "loss": 0.0566,
+      "step": 308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946382678053609e-05,
+      "loss": 0.0724,
+      "step": 309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9459352787339104e-05,
+      "loss": 0.0499,
+      "step": 310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9454860723972565e-05,
+      "loss": 0.0739,
+      "step": 311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9450350599017647e-05,
+      "loss": 0.0654,
+      "step": 312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.944582242109002e-05,
+      "loss": 0.0469,
+      "step": 313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9441276198839853e-05,
+      "loss": 0.0518,
+      "step": 314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9436711940951777e-05,
+      "loss": 0.0596,
+      "step": 315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.943212965614488e-05,
+      "loss": 0.0541,
+      "step": 316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.942752935317268e-05,
+      "loss": 0.0568,
+      "step": 317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9422911040823126e-05,
+      "loss": 0.0606,
+      "step": 318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941827472791856e-05,
+      "loss": 0.0628,
+      "step": 319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.941362042331572e-05,
+      "loss": 0.0492,
+      "step": 320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9408948135905702e-05,
+      "loss": 0.0495,
+      "step": 321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.940425787461397e-05,
+      "loss": 0.0482,
+      "step": 322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.93995496484003e-05,
+      "loss": 0.0515,
+      "step": 323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9394823466258816e-05,
+      "loss": 0.05,
+      "step": 324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9390079337217916e-05,
+      "loss": 0.0546,
+      "step": 325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9385317270340298e-05,
+      "loss": 0.0581,
+      "step": 326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9380537274722925e-05,
+      "loss": 0.0654,
+      "step": 327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9375739359497e-05,
+      "loss": 0.0343,
+      "step": 328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.937092353382797e-05,
+      "loss": 0.053,
+      "step": 329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9366089806915492e-05,
+      "loss": 0.0463,
+      "step": 330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.936123818799341e-05,
+      "loss": 0.0638,
+      "step": 331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9356368686329766e-05,
+      "loss": 0.0522,
+      "step": 332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.935148131122674e-05,
+      "loss": 0.0474,
+      "step": 333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9346576072020674e-05,
+      "loss": 0.0471,
+      "step": 334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9341652978082033e-05,
+      "loss": 0.0639,
+      "step": 335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9336712038815383e-05,
+      "loss": 0.0245,
+      "step": 336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9331753263659386e-05,
+      "loss": 0.0538,
+      "step": 337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932677666208677e-05,
+      "loss": 0.0527,
+      "step": 338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.932178224360432e-05,
+      "loss": 0.0495,
+      "step": 339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.931677001775286e-05,
+      "loss": 0.0493,
+      "step": 340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9311739994107232e-05,
+      "loss": 0.0403,
+      "step": 341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.930669218227626e-05,
+      "loss": 0.0573,
+      "step": 342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9301626591902774e-05,
+      "loss": 0.0604,
+      "step": 343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9296543232663544e-05,
+      "loss": 0.0534,
+      "step": 344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9291442114269306e-05,
+      "loss": 0.0499,
+      "step": 345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.92863232464647e-05,
+      "loss": 0.0486,
+      "step": 346
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9281186639028285e-05,
+      "loss": 0.0577,
+      "step": 347
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9276032301772503e-05,
+      "loss": 0.0538,
+      "step": 348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9270860244543666e-05,
+      "loss": 0.0505,
+      "step": 349
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9265670477221946e-05,
+      "loss": 0.0443,
+      "step": 350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9260463009721328e-05,
+      "loss": 0.043,
+      "step": 351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9255237851989624e-05,
+      "loss": 0.0716,
+      "step": 352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9249995014008434e-05,
+      "loss": 0.0571,
+      "step": 353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9244734505793136e-05,
+      "loss": 0.0476,
+      "step": 354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9239456337392854e-05,
+      "loss": 0.0445,
+      "step": 355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9234160518890463e-05,
+      "loss": 0.0428,
+      "step": 356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9228847060402543e-05,
+      "loss": 0.0536,
+      "step": 357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.922351597207938e-05,
+      "loss": 0.0448,
+      "step": 358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9218167264104925e-05,
+      "loss": 0.0498,
+      "step": 359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9212800946696806e-05,
+      "loss": 0.0632,
+      "step": 360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9207417030106277e-05,
+      "loss": 0.0568,
+      "step": 361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9202015524618216e-05,
+      "loss": 0.0599,
+      "step": 362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.91965964405511e-05,
+      "loss": 0.0551,
+      "step": 363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.919115978825699e-05,
+      "loss": 0.0532,
+      "step": 364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.91857055781215e-05,
+      "loss": 0.0547,
+      "step": 365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.918023382056379e-05,
+      "loss": 0.0629,
+      "step": 366
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9174744526036546e-05,
+      "loss": 0.06,
+      "step": 367
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.916923770502594e-05,
+      "loss": 0.0609,
+      "step": 368
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9163713368051638e-05,
+      "loss": 0.0549,
+      "step": 369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.915817152566676e-05,
+      "loss": 0.0706,
+      "step": 370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9152612188457875e-05,
+      "loss": 0.054,
+      "step": 371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9147035367044954e-05,
+      "loss": 0.0627,
+      "step": 372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9141441072081394e-05,
+      "loss": 0.0427,
+      "step": 373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9135829314253946e-05,
+      "loss": 0.0462,
+      "step": 374
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.913020010428274e-05,
+      "loss": 0.0576,
+      "step": 375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9124553452921234e-05,
+      "loss": 0.0509,
+      "step": 376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9118889370956206e-05,
+      "loss": 0.0421,
+      "step": 377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9113207869207732e-05,
+      "loss": 0.0661,
+      "step": 378
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9107508958529166e-05,
+      "loss": 0.0536,
+      "step": 379
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.910179264980712e-05,
+      "loss": 0.0465,
+      "step": 380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9096058953961438e-05,
+      "loss": 0.0732,
+      "step": 381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9090307881945177e-05,
+      "loss": 0.04,
+      "step": 382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9084539444744597e-05,
+      "loss": 0.0445,
+      "step": 383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9078753653379117e-05,
+      "loss": 0.0382,
+      "step": 384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9072950518901316e-05,
+      "loss": 0.0478,
+      "step": 385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9067130052396904e-05,
+      "loss": 0.0499,
+      "step": 386
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.90612922649847e-05,
+      "loss": 0.0573,
+      "step": 387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9055437167816605e-05,
+      "loss": 0.0571,
+      "step": 388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9049564772077595e-05,
+      "loss": 0.0564,
+      "step": 389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.904367508898569e-05,
+      "loss": 0.0514,
+      "step": 390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9037768129791924e-05,
+      "loss": 0.0594,
+      "step": 391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9031843905780344e-05,
+      "loss": 0.0542,
+      "step": 392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.902590242826798e-05,
+      "loss": 0.0462,
+      "step": 393
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9019943708604807e-05,
+      "loss": 0.0719,
+      "step": 394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9013967758173754e-05,
+      "loss": 0.0488,
+      "step": 395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.900797458839066e-05,
+      "loss": 0.0619,
+      "step": 396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9001964210704254e-05,
+      "loss": 0.0497,
+      "step": 397
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.899593663659614e-05,
+      "loss": 0.068,
+      "step": 398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.898989187758077e-05,
+      "loss": 0.0452,
+      "step": 399
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8983829945205433e-05,
+      "loss": 0.048,
+      "step": 400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.897775085105021e-05,
+      "loss": 0.0727,
+      "step": 401
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8971654606727983e-05,
+      "loss": 0.0462,
+      "step": 402
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.896554122388438e-05,
+      "loss": 0.0451,
+      "step": 403
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8959410714197775e-05,
+      "loss": 0.0684,
+      "step": 404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8953263089379265e-05,
+      "loss": 0.0662,
+      "step": 405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8947098361172634e-05,
+      "loss": 0.0689,
+      "step": 406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8940916541354343e-05,
+      "loss": 0.0516,
+      "step": 407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.89347176417335e-05,
+      "loss": 0.0598,
+      "step": 408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.892850167415184e-05,
+      "loss": 0.0499,
+      "step": 409
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8922268650483705e-05,
+      "loss": 0.0616,
+      "step": 410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8916018582636027e-05,
+      "loss": 0.0554,
+      "step": 411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.890975148254828e-05,
+      "loss": 0.0535,
+      "step": 412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8903467362192483e-05,
+      "loss": 0.0482,
+      "step": 413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.889716623357318e-05,
+      "loss": 0.0629,
+      "step": 414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8890848108727385e-05,
+      "loss": 0.066,
+      "step": 415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8884512999724596e-05,
+      "loss": 0.068,
+      "step": 416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8878160918666756e-05,
+      "loss": 0.0561,
+      "step": 417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.887179187768821e-05,
+      "loss": 0.0522,
+      "step": 418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8865405888955726e-05,
+      "loss": 0.0629,
+      "step": 419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.885900296466843e-05,
+      "loss": 0.0653,
+      "step": 420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8852583117057815e-05,
+      "loss": 0.0466,
+      "step": 421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8846146358387685e-05,
+      "loss": 0.0409,
+      "step": 422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8839692700954164e-05,
+      "loss": 0.0392,
+      "step": 423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.883322215708565e-05,
+      "loss": 0.0504,
+      "step": 424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.88267347391428e-05,
+      "loss": 0.0508,
+      "step": 425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.882023045951851e-05,
+      "loss": 0.0428,
+      "step": 426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8813709330637878e-05,
+      "loss": 0.0504,
+      "step": 427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.88071713649582e-05,
+      "loss": 0.0524,
+      "step": 428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8800616574968927e-05,
+      "loss": 0.0504,
+      "step": 429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.879404497319165e-05,
+      "loss": 0.0584,
+      "step": 430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.878745657218008e-05,
+      "loss": 0.0459,
+      "step": 431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.878085138452002e-05,
+      "loss": 0.0527,
+      "step": 432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.877422942282933e-05,
+      "loss": 0.0454,
+      "step": 433
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876759069975793e-05,
+      "loss": 0.0484,
+      "step": 434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.876093522798774e-05,
+      "loss": 0.0584,
+      "step": 435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8754263020232696e-05,
+      "loss": 0.0564,
+      "step": 436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8747574089238684e-05,
+      "loss": 0.0432,
+      "step": 437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8740868447783555e-05,
+      "loss": 0.0464,
+      "step": 438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8734146108677068e-05,
+      "loss": 0.065,
+      "step": 439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872740708476088e-05,
+      "loss": 0.0614,
+      "step": 440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8720651388908538e-05,
+      "loss": 0.0633,
+      "step": 441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8713879034025418e-05,
+      "loss": 0.0463,
+      "step": 442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8707090033048724e-05,
+      "loss": 0.0484,
+      "step": 443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.870028439894746e-05,
+      "loss": 0.0597,
+      "step": 444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8693462144722415e-05,
+      "loss": 0.0515,
+      "step": 445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8686623283406112e-05,
+      "loss": 0.0532,
+      "step": 446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8679767828062804e-05,
+      "loss": 0.0582,
+      "step": 447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8672895791788447e-05,
+      "loss": 0.0471,
+      "step": 448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866600718771067e-05,
+      "loss": 0.0662,
+      "step": 449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.865910202898874e-05,
+      "loss": 0.0515,
+      "step": 450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8652180328813575e-05,
+      "loss": 0.0614,
+      "step": 451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8645242100407655e-05,
+      "loss": 0.0543,
+      "step": 452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8638287357025074e-05,
+      "loss": 0.046,
+      "step": 453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.863131611195144e-05,
+      "loss": 0.0546,
+      "step": 454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.862432837850391e-05,
+      "loss": 0.0513,
+      "step": 455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8617324170031118e-05,
+      "loss": 0.052,
+      "step": 456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8610303499913182e-05,
+      "loss": 0.0568,
+      "step": 457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.860326638156167e-05,
+      "loss": 0.0383,
+      "step": 458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.859621282841956e-05,
+      "loss": 0.0626,
+      "step": 459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8589142853961232e-05,
+      "loss": 0.0606,
+      "step": 460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8582056471692437e-05,
+      "loss": 0.0538,
+      "step": 461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857495369515026e-05,
+      "loss": 0.0521,
+      "step": 462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8567834537903118e-05,
+      "loss": 0.0523,
+      "step": 463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.856069901355071e-05,
+      "loss": 0.0417,
+      "step": 464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8553547135724002e-05,
+      "loss": 0.0457,
+      "step": 465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8546378918085207e-05,
+      "loss": 0.0593,
+      "step": 466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8539194374327743e-05,
+      "loss": 0.0559,
+      "step": 467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.853199351817622e-05,
+      "loss": 0.0671,
+      "step": 468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.852477636338641e-05,
+      "loss": 0.0387,
+      "step": 469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8517542923745216e-05,
+      "loss": 0.053,
+      "step": 470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8510293213070662e-05,
+      "loss": 0.0444,
+      "step": 471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8503027245211836e-05,
+      "loss": 0.0509,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8495745034048898e-05,
+      "loss": 0.0494,
+      "step": 473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8488446593493028e-05,
+      "loss": 0.0729,
+      "step": 474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8481131937486413e-05,
+      "loss": 0.0525,
+      "step": 475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8473801080002216e-05,
+      "loss": 0.0436,
+      "step": 476
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.846645403504455e-05,
+      "loss": 0.0531,
+      "step": 477
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8459090816648446e-05,
+      "loss": 0.0465,
+      "step": 478
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.845171143887984e-05,
+      "loss": 0.0555,
+      "step": 479
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8444315915835533e-05,
+      "loss": 0.0674,
+      "step": 480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8436904261643162e-05,
+      "loss": 0.0564,
+      "step": 481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.842947649046119e-05,
+      "loss": 0.047,
+      "step": 482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.842203261647886e-05,
+      "loss": 0.0475,
+      "step": 483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.841457265391618e-05,
+      "loss": 0.0509,
+      "step": 484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.840709661702389e-05,
+      "loss": 0.0531,
+      "step": 485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8399604520083444e-05,
+      "loss": 0.0751,
+      "step": 486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.839209637740696e-05,
+      "loss": 0.0425,
+      "step": 487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8384572203337225e-05,
+      "loss": 0.0544,
+      "step": 488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.837703201224764e-05,
+      "loss": 0.0414,
+      "step": 489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8369475818542207e-05,
+      "loss": 0.0672,
+      "step": 490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8361903636655495e-05,
+      "loss": 0.0535,
+      "step": 491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8354315481052626e-05,
+      "loss": 0.049,
+      "step": 492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8346711366229218e-05,
+      "loss": 0.0464,
+      "step": 493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8339091306711386e-05,
+      "loss": 0.0609,
+      "step": 494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8331455317055707e-05,
+      "loss": 0.0586,
+      "step": 495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832380341184919e-05,
+      "loss": 0.0509,
+      "step": 496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831613560570924e-05,
+      "loss": 0.0495,
+      "step": 497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.830845191328364e-05,
+      "loss": 0.0522,
+      "step": 498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8300752349250524e-05,
+      "loss": 0.0412,
+      "step": 499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8293036928318336e-05,
+      "loss": 0.0505,
+      "step": 500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.828530566522583e-05,
+      "loss": 0.0464,
+      "step": 501
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8277558574742003e-05,
+      "loss": 0.0409,
+      "step": 502
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.82697956716661e-05,
+      "loss": 0.0498,
+      "step": 503
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8262016970827567e-05,
+      "loss": 0.0593,
+      "step": 504
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8254222487086028e-05,
+      "loss": 0.036,
+      "step": 505
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.824641223533126e-05,
+      "loss": 0.0592,
+      "step": 506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.823858623048316e-05,
+      "loss": 0.0516,
+      "step": 507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8230744487491722e-05,
+      "loss": 0.0464,
+      "step": 508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8222887021337003e-05,
+      "loss": 0.0611,
+      "step": 509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.821501384702908e-05,
+      "loss": 0.0443,
+      "step": 510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8207124979608067e-05,
+      "loss": 0.058,
+      "step": 511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.819922043414403e-05,
+      "loss": 0.0557,
+      "step": 512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8191300225737004e-05,
+      "loss": 0.0344,
+      "step": 513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8183364369516928e-05,
+      "loss": 0.046,
+      "step": 514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8175412880643647e-05,
+      "loss": 0.0587,
+      "step": 515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8167445774306866e-05,
+      "loss": 0.0284,
+      "step": 516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8159463065726108e-05,
+      "loss": 0.043,
+      "step": 517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.815146477015073e-05,
+      "loss": 0.0484,
+      "step": 518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.814345090285984e-05,
+      "loss": 0.0526,
+      "step": 519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.81354214791623e-05,
+      "loss": 0.0429,
+      "step": 520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8127376514396698e-05,
+      "loss": 0.0555,
+      "step": 521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8119316023931298e-05,
+      "loss": 0.0484,
+      "step": 522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8111240023164027e-05,
+      "loss": 0.0562,
+      "step": 523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8103148527522444e-05,
+      "loss": 0.0483,
+      "step": 524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8095041552463705e-05,
+      "loss": 0.0512,
+      "step": 525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8086919113474537e-05,
+      "loss": 0.0529,
+      "step": 526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8078781226071208e-05,
+      "loss": 0.0408,
+      "step": 527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8070627905799497e-05,
+      "loss": 0.0508,
+      "step": 528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8062459168234665e-05,
+      "loss": 0.0543,
+      "step": 529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.805427502898142e-05,
+      "loss": 0.0455,
+      "step": 530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.80460755036739e-05,
+      "loss": 0.043,
+      "step": 531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8037860607975627e-05,
+      "loss": 0.0526,
+      "step": 532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.802963035757949e-05,
+      "loss": 0.0383,
+      "step": 533
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8021384768207705e-05,
+      "loss": 0.0418,
+      "step": 534
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8013123855611797e-05,
+      "loss": 0.0609,
+      "step": 535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.800484763557256e-05,
+      "loss": 0.0337,
+      "step": 536
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7996556123900032e-05,
+      "loss": 0.0545,
+      "step": 537
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.798824933643345e-05,
+      "loss": 0.0485,
+      "step": 538
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.797992728904125e-05,
+      "loss": 0.0529,
+      "step": 539
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7971589997621005e-05,
+      "loss": 0.0304,
+      "step": 540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.796323747809942e-05,
+      "loss": 0.0449,
+      "step": 541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7954869746432282e-05,
+      "loss": 0.0366,
+      "step": 542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.794648681860444e-05,
+      "loss": 0.0466,
+      "step": 543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7938088710629778e-05,
+      "loss": 0.0497,
+      "step": 544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7929675438551163e-05,
+      "loss": 0.0425,
+      "step": 545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.792124701844045e-05,
+      "loss": 0.0464,
+      "step": 546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7912803466398415e-05,
+      "loss": 0.0559,
+      "step": 547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.790434479855475e-05,
+      "loss": 0.0595,
+      "step": 548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7895871031068016e-05,
+      "loss": 0.0468,
+      "step": 549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7887382180125627e-05,
+      "loss": 0.0473,
+      "step": 550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7878878261943803e-05,
+      "loss": 0.0504,
+      "step": 551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7870359292767552e-05,
+      "loss": 0.0462,
+      "step": 552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7861825288870633e-05,
+      "loss": 0.0622,
+      "step": 553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.785327626655552e-05,
+      "loss": 0.0544,
+      "step": 554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7844712242153382e-05,
+      "loss": 0.0438,
+      "step": 555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7836133232024053e-05,
+      "loss": 0.0547,
+      "step": 556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.782753925255598e-05,
+      "loss": 0.0395,
+      "step": 557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.781893032016621e-05,
+      "loss": 0.0489,
+      "step": 558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7810306451300368e-05,
+      "loss": 0.0426,
+      "step": 559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7801667662432595e-05,
+      "loss": 0.0542,
+      "step": 560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.779301397006554e-05,
+      "loss": 0.0497,
+      "step": 561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.778434539073032e-05,
+      "loss": 0.0481,
+      "step": 562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7775661940986496e-05,
+      "loss": 0.0469,
+      "step": 563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7766963637422034e-05,
+      "loss": 0.0494,
+      "step": 564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7758250496653268e-05,
+      "loss": 0.0591,
+      "step": 565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7749522535324887e-05,
+      "loss": 0.0489,
+      "step": 566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7740779770109884e-05,
+      "loss": 0.0401,
+      "step": 567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7732022217709537e-05,
+      "loss": 0.0436,
+      "step": 568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7723249894853362e-05,
+      "loss": 0.054,
+      "step": 569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7714462818299103e-05,
+      "loss": 0.0437,
+      "step": 570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7705661004832678e-05,
+      "loss": 0.0579,
+      "step": 571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7696844471268163e-05,
+      "loss": 0.0518,
+      "step": 572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.768801323444776e-05,
+      "loss": 0.0525,
+      "step": 573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.767916731124174e-05,
+      "loss": 0.0523,
+      "step": 574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.767030671854844e-05,
+      "loss": 0.0658,
+      "step": 575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.766143147329423e-05,
+      "loss": 0.0465,
+      "step": 576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.765254159243345e-05,
+      "loss": 0.0436,
+      "step": 577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7643637092948418e-05,
+      "loss": 0.0453,
+      "step": 578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7634717991849366e-05,
+      "loss": 0.0516,
+      "step": 579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7625784306174417e-05,
+      "loss": 0.0454,
+      "step": 580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7616836052989574e-05,
+      "loss": 0.0543,
+      "step": 581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.760787324938864e-05,
+      "loss": 0.0558,
+      "step": 582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7598895912493234e-05,
+      "loss": 0.0555,
+      "step": 583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7589904059452734e-05,
+      "loss": 0.0669,
+      "step": 584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.758089770744425e-05,
+      "loss": 0.0326,
+      "step": 585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7571876873672575e-05,
+      "loss": 0.039,
+      "step": 586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7562841575370185e-05,
+      "loss": 0.0435,
+      "step": 587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.755379182979718e-05,
+      "loss": 0.0548,
+      "step": 588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7544727654241254e-05,
+      "loss": 0.0409,
+      "step": 589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.753564906601767e-05,
+      "loss": 0.0403,
+      "step": 590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7526556082469234e-05,
+      "loss": 0.0507,
+      "step": 591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.751744872096623e-05,
+      "loss": 0.0558,
+      "step": 592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7508326998906423e-05,
+      "loss": 0.047,
+      "step": 593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7499190933715015e-05,
+      "loss": 0.0576,
+      "step": 594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.749004054284459e-05,
+      "loss": 0.0565,
+      "step": 595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7480875843775115e-05,
+      "loss": 0.0526,
+      "step": 596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.747169685401388e-05,
+      "loss": 0.042,
+      "step": 597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7462503591095487e-05,
+      "loss": 0.0429,
+      "step": 598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.745329607258178e-05,
+      "loss": 0.0487,
+      "step": 599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7444074316061864e-05,
+      "loss": 0.0455,
+      "step": 600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7434838339152018e-05,
+      "loss": 0.0491,
+      "step": 601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7425588159495706e-05,
+      "loss": 0.0366,
+      "step": 602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7416323794763514e-05,
+      "loss": 0.0564,
+      "step": 603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.740704526265312e-05,
+      "loss": 0.0404,
+      "step": 604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.739775258088928e-05,
+      "loss": 0.0444,
+      "step": 605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7388445767223766e-05,
+      "loss": 0.0399,
+      "step": 606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7379124839435356e-05,
+      "loss": 0.0486,
+      "step": 607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7369789815329792e-05,
+      "loss": 0.0553,
+      "step": 608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.736044071273973e-05,
+      "loss": 0.0513,
+      "step": 609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7351077549524728e-05,
+      "loss": 0.038,
+      "step": 610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7341700343571214e-05,
+      "loss": 0.0574,
+      "step": 611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7332309112792427e-05,
+      "loss": 0.0518,
+      "step": 612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7322903875128405e-05,
+      "loss": 0.0584,
+      "step": 613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.731348464854594e-05,
+      "loss": 0.042,
+      "step": 614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.730405145103855e-05,
+      "loss": 0.0463,
+      "step": 615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.729460430062644e-05,
+      "loss": 0.0685,
+      "step": 616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.728514321535647e-05,
+      "loss": 0.0417,
+      "step": 617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7275668213302117e-05,
+      "loss": 0.0576,
+      "step": 618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.726617931256345e-05,
+      "loss": 0.0457,
+      "step": 619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7256676531267075e-05,
+      "loss": 0.051,
+      "step": 620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.724715988756613e-05,
+      "loss": 0.044,
+      "step": 621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7237629399640226e-05,
+      "loss": 0.0515,
+      "step": 622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7228085085695422e-05,
+      "loss": 0.0558,
+      "step": 623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7218526963964188e-05,
+      "loss": 0.0415,
+      "step": 624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7208955052705373e-05,
+      "loss": 0.0568,
+      "step": 625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.719936937020416e-05,
+      "loss": 0.0487,
+      "step": 626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7189769934772057e-05,
+      "loss": 0.0348,
+      "step": 627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7180156764746825e-05,
+      "loss": 0.0534,
+      "step": 628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.717052987849247e-05,
+      "loss": 0.0449,
+      "step": 629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7160889294399203e-05,
+      "loss": 0.0409,
+      "step": 630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.71512350308834e-05,
+      "loss": 0.0517,
+      "step": 631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.714156710638756e-05,
+      "loss": 0.0513,
+      "step": 632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.71318855393803e-05,
+      "loss": 0.0535,
+      "step": 633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.712219034835627e-05,
+      "loss": 0.0457,
+      "step": 634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7112481551836175e-05,
+      "loss": 0.0439,
+      "step": 635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7102759168366683e-05,
+      "loss": 0.0489,
+      "step": 636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.709302321652044e-05,
+      "loss": 0.0448,
+      "step": 637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7083273714895993e-05,
+      "loss": 0.0486,
+      "step": 638
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.707351068211779e-05,
+      "loss": 0.0501,
+      "step": 639
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7063734136836116e-05,
+      "loss": 0.0655,
+      "step": 640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.705394409772707e-05,
+      "loss": 0.0464,
+      "step": 641
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7044140583492535e-05,
+      "loss": 0.0612,
+      "step": 642
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7034323612860125e-05,
+      "loss": 0.0532,
+      "step": 643
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7024493204583173e-05,
+      "loss": 0.0503,
+      "step": 644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7014649377440672e-05,
+      "loss": 0.0408,
+      "step": 645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.700479215023725e-05,
+      "loss": 0.05,
+      "step": 646
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6994921541803134e-05,
+      "loss": 0.0417,
+      "step": 647
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6985037570994114e-05,
+      "loss": 0.0482,
+      "step": 648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.697514025669151e-05,
+      "loss": 0.0513,
+      "step": 649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6965229617802122e-05,
+      "loss": 0.0485,
+      "step": 650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6955305673258213e-05,
+      "loss": 0.0422,
+      "step": 651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6945368442017463e-05,
+      "loss": 0.0656,
+      "step": 652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.693541794306293e-05,
+      "loss": 0.0429,
+      "step": 653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6925454195403013e-05,
+      "loss": 0.0532,
+      "step": 654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6915477218071432e-05,
+      "loss": 0.0594,
+      "step": 655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6905487030127173e-05,
+      "loss": 0.0519,
+      "step": 656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.689548365065445e-05,
+      "loss": 0.0514,
+      "step": 657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6885467098762692e-05,
+      "loss": 0.055,
+      "step": 658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6875437393586485e-05,
+      "loss": 0.0491,
+      "step": 659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.686539455428554e-05,
+      "loss": 0.0511,
+      "step": 660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.685533860004465e-05,
+      "loss": 0.0592,
+      "step": 661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6845269550073684e-05,
+      "loss": 0.0436,
+      "step": 662
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6835187423607505e-05,
+      "loss": 0.0473,
+      "step": 663
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.682509223990597e-05,
+      "loss": 0.0406,
+      "step": 664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.681498401825387e-05,
+      "loss": 0.0371,
+      "step": 665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.680486277796091e-05,
+      "loss": 0.045,
+      "step": 666
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6794728538361656e-05,
+      "loss": 0.036,
+      "step": 667
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6784581318815517e-05,
+      "loss": 0.0368,
+      "step": 668
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6774421138706687e-05,
+      "loss": 0.0493,
+      "step": 669
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6764248017444128e-05,
+      "loss": 0.0486,
+      "step": 670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.675406197446152e-05,
+      "loss": 0.0395,
+      "step": 671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6743863029217222e-05,
+      "loss": 0.0482,
+      "step": 672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6733651201194247e-05,
+      "loss": 0.053,
+      "step": 673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6723426509900215e-05,
+      "loss": 0.0551,
+      "step": 674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6713188974867317e-05,
+      "loss": 0.0348,
+      "step": 675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6702938615652287e-05,
+      "loss": 0.0447,
+      "step": 676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6692675451836348e-05,
+      "loss": 0.0347,
+      "step": 677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6682399503025185e-05,
+      "loss": 0.0711,
+      "step": 678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6672110788848912e-05,
+      "loss": 0.0541,
+      "step": 679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.666180932896202e-05,
+      "loss": 0.0502,
+      "step": 680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6651495143043358e-05,
+      "loss": 0.0463,
+      "step": 681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.664116825079608e-05,
+      "loss": 0.0438,
+      "step": 682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663082867194761e-05,
+      "loss": 0.042,
+      "step": 683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6620476426249607e-05,
+      "loss": 0.0534,
+      "step": 684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661011153347794e-05,
+      "loss": 0.0395,
+      "step": 685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6599734013432627e-05,
+      "loss": 0.0488,
+      "step": 686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6589343885937802e-05,
+      "loss": 0.0505,
+      "step": 687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6578941170841697e-05,
+      "loss": 0.0451,
+      "step": 688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6568525888016583e-05,
+      "loss": 0.0409,
+      "step": 689
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6558098057358735e-05,
+      "loss": 0.0515,
+      "step": 690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6547657698788405e-05,
+      "loss": 0.0503,
+      "step": 691
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.653720483224978e-05,
+      "loss": 0.044,
+      "step": 692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6526739477710925e-05,
+      "loss": 0.0587,
+      "step": 693
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6516261655163776e-05,
+      "loss": 0.032,
+      "step": 694
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6505771384624086e-05,
+      "loss": 0.0375,
+      "step": 695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.649526868613138e-05,
+      "loss": 0.051,
+      "step": 696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6484753579748926e-05,
+      "loss": 0.0405,
+      "step": 697
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6474226085563697e-05,
+      "loss": 0.0427,
+      "step": 698
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.646368622368633e-05,
+      "loss": 0.0438,
+      "step": 699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6453134014251085e-05,
+      "loss": 0.0467,
+      "step": 700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.644256947741582e-05,
+      "loss": 0.0516,
+      "step": 701
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6431992633361923e-05,
+      "loss": 0.0424,
+      "step": 702
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.642140350229431e-05,
+      "loss": 0.0418,
+      "step": 703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6410802104441355e-05,
+      "loss": 0.0471,
+      "step": 704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6400188460054886e-05,
+      "loss": 0.0388,
+      "step": 705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6389562589410098e-05,
+      "loss": 0.0448,
+      "step": 706
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6378924512805566e-05,
+      "loss": 0.0515,
+      "step": 707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6368274250563163e-05,
+      "loss": 0.0577,
+      "step": 708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6357611823028055e-05,
+      "loss": 0.0379,
+      "step": 709
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6346937250568638e-05,
+      "loss": 0.0509,
+      "step": 710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6336250553576514e-05,
+      "loss": 0.0444,
+      "step": 711
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.632555175246644e-05,
+      "loss": 0.0549,
+      "step": 712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6314840867676303e-05,
+      "loss": 0.0446,
+      "step": 713
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6304117919667068e-05,
+      "loss": 0.0432,
+      "step": 714
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.629338292892275e-05,
+      "loss": 0.049,
+      "step": 715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.628263591595036e-05,
+      "loss": 0.0486,
+      "step": 716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6271876901279885e-05,
+      "loss": 0.0463,
+      "step": 717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.626110590546423e-05,
+      "loss": 0.0566,
+      "step": 718
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6250322949079197e-05,
+      "loss": 0.0435,
+      "step": 719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6239528052723426e-05,
+      "loss": 0.0411,
+      "step": 720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6228721237018376e-05,
+      "loss": 0.0586,
+      "step": 721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6217902522608264e-05,
+      "loss": 0.0584,
+      "step": 722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6207071930160046e-05,
+      "loss": 0.0399,
+      "step": 723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6196229480363367e-05,
+      "loss": 0.0423,
+      "step": 724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6185375193930525e-05,
+      "loss": 0.048,
+      "step": 725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6174509091596422e-05,
+      "loss": 0.0457,
+      "step": 726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.616363119411854e-05,
+      "loss": 0.0426,
+      "step": 727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6152741522276884e-05,
+      "loss": 0.0445,
+      "step": 728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6141840096873963e-05,
+      "loss": 0.04,
+      "step": 729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.613092693873473e-05,
+      "loss": 0.0572,
+      "step": 730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6120002068706556e-05,
+      "loss": 0.0427,
+      "step": 731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6109065507659184e-05,
+      "loss": 0.0346,
+      "step": 732
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.609811727648468e-05,
+      "loss": 0.0318,
+      "step": 733
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6087157396097427e-05,
+      "loss": 0.0433,
+      "step": 734
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.607618588743404e-05,
+      "loss": 0.0572,
+      "step": 735
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.606520277145335e-05,
+      "loss": 0.041,
+      "step": 736
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.605420806913638e-05,
+      "loss": 0.0522,
+      "step": 737
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.604320180148626e-05,
+      "loss": 0.0438,
+      "step": 738
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6032183989528228e-05,
+      "loss": 0.0498,
+      "step": 739
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.602115465430958e-05,
+      "loss": 0.0502,
+      "step": 740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6010113816899608e-05,
+      "loss": 0.0357,
+      "step": 741
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5999061498389593e-05,
+      "loss": 0.0531,
+      "step": 742
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5987997719892735e-05,
+      "loss": 0.0519,
+      "step": 743
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5976922502544142e-05,
+      "loss": 0.0413,
+      "step": 744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5965835867500758e-05,
+      "loss": 0.0507,
+      "step": 745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.595473783594134e-05,
+      "loss": 0.0526,
+      "step": 746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5943628429066434e-05,
+      "loss": 0.0721,
+      "step": 747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.593250766809829e-05,
+      "loss": 0.0524,
+      "step": 748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5921375574280865e-05,
+      "loss": 0.044,
+      "step": 749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.591023216887976e-05,
+      "loss": 0.045,
+      "step": 750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5899077473182178e-05,
+      "loss": 0.0612,
+      "step": 751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5887911508496905e-05,
+      "loss": 0.0394,
+      "step": 752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.587673429615424e-05,
+      "loss": 0.0524,
+      "step": 753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.586554585750598e-05,
+      "loss": 0.0439,
+      "step": 754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.5854346213925354e-05,
+      "loss": 0.053,
+      "step": 755
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.584313538680701e-05,
+      "loss": 0.052,
+      "step": 756
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5831913397566947e-05,
+      "loss": 0.0425,
+      "step": 757
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5820680267642497e-05,
+      "loss": 0.0534,
+      "step": 758
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5809436018492267e-05,
+      "loss": 0.0422,
+      "step": 759
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.579818067159611e-05,
+      "loss": 0.0508,
+      "step": 760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5786914248455077e-05,
+      "loss": 0.0363,
+      "step": 761
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5775636770591375e-05,
+      "loss": 0.036,
+      "step": 762
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5764348259548336e-05,
+      "loss": 0.0385,
+      "step": 763
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5753048736890364e-05,
+      "loss": 0.0458,
+      "step": 764
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.5741738224202895e-05,
+      "loss": 0.0453,
+      "step": 765
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.573041674309237e-05,
+      "loss": 0.0593,
+      "step": 766
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.571908431518617e-05,
+      "loss": 0.0552,
+      "step": 767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5707740962132593e-05,
+      "loss": 0.0653,
+      "step": 768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.569638670560081e-05,
+      "loss": 0.0579,
+      "step": 769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5685021567280818e-05,
+      "loss": 0.0448,
+      "step": 770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5673645568883406e-05,
+      "loss": 0.0408,
+      "step": 771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.56622587321401e-05,
+      "loss": 0.0516,
+      "step": 772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.565086107880314e-05,
+      "loss": 0.0507,
+      "step": 773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5639452630645418e-05,
+      "loss": 0.0555,
+      "step": 774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.562803340946046e-05,
+      "loss": 0.0367,
+      "step": 775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5616603437062364e-05,
+      "loss": 0.0375,
+      "step": 776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5605162735285768e-05,
+      "loss": 0.0393,
+      "step": 777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.5593711325985803e-05,
+      "loss": 0.0524,
+      "step": 778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.558224923103806e-05,
+      "loss": 0.0517,
+      "step": 779
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.557077647233854e-05,
+      "loss": 0.0569,
+      "step": 780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5559293071803616e-05,
+      "loss": 0.0618,
+      "step": 781
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.554779905136998e-05,
+      "loss": 0.0429,
+      "step": 782
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.553629443299464e-05,
+      "loss": 0.0563,
+      "step": 783
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.552477923865481e-05,
+      "loss": 0.0509,
+      "step": 784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.551325349034793e-05,
+      "loss": 0.0459,
+      "step": 785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5501717210091605e-05,
+      "loss": 0.0362,
+      "step": 786
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5490170419923554e-05,
+      "loss": 0.0403,
+      "step": 787
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.547861314190156e-05,
+      "loss": 0.0468,
+      "step": 788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.546704539810346e-05,
+      "loss": 0.0527,
+      "step": 789
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5455467210627073e-05,
+      "loss": 0.0392,
+      "step": 790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5443878601590175e-05,
+      "loss": 0.0519,
+      "step": 791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.543227959313044e-05,
+      "loss": 0.0453,
+      "step": 792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.542067020740542e-05,
+      "loss": 0.0418,
+      "step": 793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5409050466592484e-05,
+      "loss": 0.0445,
+      "step": 794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.539742039288878e-05,
+      "loss": 0.0341,
+      "step": 795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5385780008511203e-05,
+      "loss": 0.0533,
+      "step": 796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5374129335696337e-05,
+      "loss": 0.0407,
+      "step": 797
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5362468396700427e-05,
+      "loss": 0.0432,
+      "step": 798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5350797213799316e-05,
+      "loss": 0.0388,
+      "step": 799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5339115809288435e-05,
+      "loss": 0.0372,
+      "step": 800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5327424205482726e-05,
+      "loss": 0.0468,
+      "step": 801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5315722424716624e-05,
+      "loss": 0.0336,
+      "step": 802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5304010489343997e-05,
+      "loss": 0.0566,
+      "step": 803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.529228842173812e-05,
+      "loss": 0.0489,
+      "step": 804
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.528055624429161e-05,
+      "loss": 0.0492,
+      "step": 805
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.526881397941642e-05,
+      "loss": 0.0486,
+      "step": 806
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5257061649543744e-05,
+      "loss": 0.0372,
+      "step": 807
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5245299277124029e-05,
+      "loss": 0.0553,
+      "step": 808
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5233526884626884e-05,
+      "loss": 0.0331,
+      "step": 809
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5221744494541085e-05,
+      "loss": 0.0412,
+      "step": 810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.520995212937448e-05,
+      "loss": 0.0615,
+      "step": 811
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5198149811653989e-05,
+      "loss": 0.0475,
+      "step": 812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5186337563925539e-05,
+      "loss": 0.0486,
+      "step": 813
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5174515408754027e-05,
+      "loss": 0.0419,
+      "step": 814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5162683368723274e-05,
+      "loss": 0.0331,
+      "step": 815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.515084146643599e-05,
+      "loss": 0.0513,
+      "step": 816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5138989724513718e-05,
+      "loss": 0.0412,
+      "step": 817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5127128165596796e-05,
+      "loss": 0.042,
+      "step": 818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.511525681234433e-05,
+      "loss": 0.0446,
+      "step": 819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5103375687434117e-05,
+      "loss": 0.0369,
+      "step": 820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.509148481356263e-05,
+      "loss": 0.0474,
+      "step": 821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5079584213444976e-05,
+      "loss": 0.0473,
+      "step": 822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.506767390981482e-05,
+      "loss": 0.0643,
+      "step": 823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5055753925424375e-05,
+      "loss": 0.0652,
+      "step": 824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5043824283044352e-05,
+      "loss": 0.0422,
+      "step": 825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5031885005463904e-05,
+      "loss": 0.0427,
+      "step": 826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5019936115490593e-05,
+      "loss": 0.0406,
+      "step": 827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5007977635950338e-05,
+      "loss": 0.0389,
+      "step": 828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4996009589687388e-05,
+      "loss": 0.0558,
+      "step": 829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.498403199956426e-05,
+      "loss": 0.0521,
+      "step": 830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4972044888461703e-05,
+      "loss": 0.0396,
+      "step": 831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4960048279278654e-05,
+      "loss": 0.0423,
+      "step": 832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4948042194932198e-05,
+      "loss": 0.0506,
+      "step": 833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4936026658357512e-05,
+      "loss": 0.0452,
+      "step": 834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4924001692507842e-05,
+      "loss": 0.0516,
+      "step": 835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4911967320354432e-05,
+      "loss": 0.0347,
+      "step": 836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.4899923564886505e-05,
+      "loss": 0.053,
+      "step": 837
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4887870449111209e-05,
+      "loss": 0.0487,
+      "step": 838
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4875807996053566e-05,
+      "loss": 0.0441,
+      "step": 839
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4863736228756445e-05,
+      "loss": 0.0528,
+      "step": 840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4851655170280493e-05,
+      "loss": 0.0481,
+      "step": 841
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4839564843704119e-05,
+      "loss": 0.0345,
+      "step": 842
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.482746527212344e-05,
+      "loss": 0.0395,
+      "step": 843
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4815356478652217e-05,
+      "loss": 0.036,
+      "step": 844
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4803238486421844e-05,
+      "loss": 0.0511,
+      "step": 845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4791111318581277e-05,
+      "loss": 0.047,
+      "step": 846
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4778974998297e-05,
+      "loss": 0.0354,
+      "step": 847
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4766829548752992e-05,
+      "loss": 0.0448,
+      "step": 848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.4754674993150659e-05,
+      "loss": 0.0467,
+      "step": 849
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4742511354708808e-05,
+      "loss": 0.0369,
+      "step": 850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4730338656663601e-05,
+      "loss": 0.0602,
+      "step": 851
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4718156922268494e-05,
+      "loss": 0.0338,
+      "step": 852
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4705966174794217e-05,
+      "loss": 0.0523,
+      "step": 853
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4693766437528715e-05,
+      "loss": 0.0362,
+      "step": 854
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4681557733777103e-05,
+      "loss": 0.0477,
+      "step": 855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4669340086861626e-05,
+      "loss": 0.0684,
+      "step": 856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.465711352012161e-05,
+      "loss": 0.0697,
+      "step": 857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4644878056913434e-05,
+      "loss": 0.0394,
+      "step": 858
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4632633720610456e-05,
+      "loss": 0.0311,
+      "step": 859
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.4620380534602985e-05,
+      "loss": 0.0482,
+      "step": 860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.460811852229825e-05,
+      "loss": 0.0436,
+      "step": 861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4595847707120325e-05,
+      "loss": 0.0399,
+      "step": 862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4583568112510109e-05,
+      "loss": 0.0532,
+      "step": 863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4571279761925267e-05,
+      "loss": 0.0361,
+      "step": 864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4558982678840202e-05,
+      "loss": 0.0464,
+      "step": 865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.454667688674598e-05,
+      "loss": 0.0389,
+      "step": 866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.453436240915032e-05,
+      "loss": 0.038,
+      "step": 867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4522039269577523e-05,
+      "loss": 0.0286,
+      "step": 868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4509707491568446e-05,
+      "loss": 0.0429,
+      "step": 869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4497367098680437e-05,
+      "loss": 0.046,
+      "step": 870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.448501811448731e-05,
+      "loss": 0.0435,
+      "step": 871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4472660562579289e-05,
+      "loss": 0.0565,
+      "step": 872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.4460294466562959e-05,
+      "loss": 0.0305,
+      "step": 873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4447919850061234e-05,
+      "loss": 0.0716,
+      "step": 874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4435536736713299e-05,
+      "loss": 0.0536,
+      "step": 875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4423145150174578e-05,
+      "loss": 0.0431,
+      "step": 876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4410745114116674e-05,
+      "loss": 0.051,
+      "step": 877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4398336652227335e-05,
+      "loss": 0.0433,
+      "step": 878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4385919788210405e-05,
+      "loss": 0.0401,
+      "step": 879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4373494545785774e-05,
+      "loss": 0.0504,
+      "step": 880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4361060948689343e-05,
+      "loss": 0.0473,
+      "step": 881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4348619020672968e-05,
+      "loss": 0.043,
+      "step": 882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.433616878550442e-05,
+      "loss": 0.0359,
+      "step": 883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.432371026696735e-05,
+      "loss": 0.0491,
+      "step": 884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4311243488861215e-05,
+      "loss": 0.0493,
+      "step": 885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4298768475001264e-05,
+      "loss": 0.0629,
+      "step": 886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4286285249218469e-05,
+      "loss": 0.035,
+      "step": 887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4273793835359495e-05,
+      "loss": 0.0271,
+      "step": 888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4261294257286653e-05,
+      "loss": 0.0452,
+      "step": 889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4248786538877836e-05,
+      "loss": 0.0487,
+      "step": 890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4236270704026507e-05,
+      "loss": 0.0533,
+      "step": 891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4223746776641614e-05,
+      "loss": 0.0396,
+      "step": 892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4211214780647574e-05,
+      "loss": 0.0443,
+      "step": 893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4198674739984218e-05,
+      "loss": 0.0493,
+      "step": 894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4186126678606745e-05,
+      "loss": 0.0402,
+      "step": 895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4173570620485668e-05,
+      "loss": 0.0483,
+      "step": 896
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4161006589606789e-05,
+      "loss": 0.0507,
+      "step": 897
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4148434609971134e-05,
+      "loss": 0.0369,
+      "step": 898
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4135854705594905e-05,
+      "loss": 0.0379,
+      "step": 899
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.412326690050946e-05,
+      "loss": 0.0407,
+      "step": 900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4110671218761237e-05,
+      "loss": 0.0368,
+      "step": 901
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.409806768441172e-05,
+      "loss": 0.0411,
+      "step": 902
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4085456321537404e-05,
+      "loss": 0.0412,
+      "step": 903
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4072837154229726e-05,
+      "loss": 0.041,
+      "step": 904
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4060210206595046e-05,
+      "loss": 0.045,
+      "step": 905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4047575502754574e-05,
+      "loss": 0.0358,
+      "step": 906
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4034933066844346e-05,
+      "loss": 0.0462,
+      "step": 907
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.402228292301516e-05,
+      "loss": 0.0477,
+      "step": 908
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4009625095432545e-05,
+      "loss": 0.0396,
+      "step": 909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3996959608276705e-05,
+      "loss": 0.0448,
+      "step": 910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3984286485742481e-05,
+      "loss": 0.0463,
+      "step": 911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3971605752039292e-05,
+      "loss": 0.0374,
+      "step": 912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3958917431391103e-05,
+      "loss": 0.0498,
+      "step": 913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.394622154803637e-05,
+      "loss": 0.0452,
+      "step": 914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3933518126227991e-05,
+      "loss": 0.0463,
+      "step": 915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3920807190233276e-05,
+      "loss": 0.0353,
+      "step": 916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.390808876433388e-05,
+      "loss": 0.0321,
+      "step": 917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3895362872825766e-05,
+      "loss": 0.0545,
+      "step": 918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3882629540019167e-05,
+      "loss": 0.0347,
+      "step": 919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3869888790238522e-05,
+      "loss": 0.053,
+      "step": 920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3857140647822444e-05,
+      "loss": 0.0396,
+      "step": 921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3844385137123667e-05,
+      "loss": 0.0462,
+      "step": 922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3831622282508996e-05,
+      "loss": 0.0541,
+      "step": 923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3818852108359271e-05,
+      "loss": 0.053,
+      "step": 924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3806074639069315e-05,
+      "loss": 0.0599,
+      "step": 925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3793289899047877e-05,
+      "loss": 0.0374,
+      "step": 926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3780497912717613e-05,
+      "loss": 0.0533,
+      "step": 927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3767698704515e-05,
+      "loss": 0.0485,
+      "step": 928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.375489229889033e-05,
+      "loss": 0.0532,
+      "step": 929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3742078720307633e-05,
+      "loss": 0.0362,
+      "step": 930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3729257993244644e-05,
+      "loss": 0.0406,
+      "step": 931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3716430142192758e-05,
+      "loss": 0.0471,
+      "step": 932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3703595191656973e-05,
+      "loss": 0.0425,
+      "step": 933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.369075316615585e-05,
+      "loss": 0.0499,
+      "step": 934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3677904090221468e-05,
+      "loss": 0.0291,
+      "step": 935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3665047988399376e-05,
+      "loss": 0.042,
+      "step": 936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.365218488524854e-05,
+      "loss": 0.0459,
+      "step": 937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.36393148053413e-05,
+      "loss": 0.0517,
+      "step": 938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3626437773263325e-05,
+      "loss": 0.0417,
+      "step": 939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.361355381361357e-05,
+      "loss": 0.0408,
+      "step": 940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.360066295100422e-05,
+      "loss": 0.0604,
+      "step": 941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3587765210060639e-05,
+      "loss": 0.0518,
+      "step": 942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.3574860615421347e-05,
+      "loss": 0.0429,
+      "step": 943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3561949191737944e-05,
+      "loss": 0.0442,
+      "step": 944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3549030963675077e-05,
+      "loss": 0.0419,
+      "step": 945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3536105955910392e-05,
+      "loss": 0.0345,
+      "step": 946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3523174193134491e-05,
+      "loss": 0.0492,
+      "step": 947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3510235700050875e-05,
+      "loss": 0.032,
+      "step": 948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3497290501375906e-05,
+      "loss": 0.0426,
+      "step": 949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3484338621838748e-05,
+      "loss": 0.0379,
+      "step": 950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3471380086181336e-05,
+      "loss": 0.041,
+      "step": 951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3458414919158316e-05,
+      "loss": 0.037,
+      "step": 952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3445443145537002e-05,
+      "loss": 0.0389,
+      "step": 953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3432464790097333e-05,
+      "loss": 0.0326,
+      "step": 954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.3419479877631811e-05,
+      "loss": 0.0468,
+      "step": 955
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3406488432945477e-05,
+      "loss": 0.0479,
+      "step": 956
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.339349048085584e-05,
+      "loss": 0.0471,
+      "step": 957
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3380486046192842e-05,
+      "loss": 0.0436,
+      "step": 958
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3367475153798812e-05,
+      "loss": 0.0494,
+      "step": 959
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3354457828528418e-05,
+      "loss": 0.0333,
+      "step": 960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3341434095248605e-05,
+      "loss": 0.0431,
+      "step": 961
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3328403978838571e-05,
+      "loss": 0.0418,
+      "step": 962
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.33153675041897e-05,
+      "loss": 0.0367,
+      "step": 963
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3302324696205533e-05,
+      "loss": 0.0429,
+      "step": 964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3289275579801691e-05,
+      "loss": 0.0509,
+      "step": 965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.3276220179905866e-05,
+      "loss": 0.0465,
+      "step": 966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.326315852145774e-05,
+      "loss": 0.0351,
+      "step": 967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3250090629408951e-05,
+      "loss": 0.0382,
+      "step": 968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3237016528723059e-05,
+      "loss": 0.0471,
+      "step": 969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3223936244375465e-05,
+      "loss": 0.0342,
+      "step": 970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3210849801353395e-05,
+      "loss": 0.0608,
+      "step": 971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3197757224655838e-05,
+      "loss": 0.0414,
+      "step": 972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3184658539293498e-05,
+      "loss": 0.0285,
+      "step": 973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3171553770288746e-05,
+      "loss": 0.0392,
+      "step": 974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3158442942675582e-05,
+      "loss": 0.0414,
+      "step": 975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.314532608149958e-05,
+      "loss": 0.0506,
+      "step": 976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3132203211817825e-05,
+      "loss": 0.0527,
+      "step": 977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3119074358698893e-05,
+      "loss": 0.0517,
+      "step": 978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3105939547222794e-05,
+      "loss": 0.047,
+      "step": 979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3092798802480909e-05,
+      "loss": 0.0399,
+      "step": 980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3079652149575957e-05,
+      "loss": 0.0316,
+      "step": 981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3066499613621946e-05,
+      "loss": 0.0418,
+      "step": 982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.305334121974412e-05,
+      "loss": 0.0369,
+      "step": 983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3040176993078919e-05,
+      "loss": 0.0467,
+      "step": 984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3027006958773914e-05,
+      "loss": 0.0348,
+      "step": 985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3013831141987781e-05,
+      "loss": 0.0451,
+      "step": 986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3000649567890235e-05,
+      "loss": 0.0389,
+      "step": 987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2987462261661996e-05,
+      "loss": 0.0502,
+      "step": 988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2974269248494729e-05,
+      "loss": 0.0245,
+      "step": 989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2961070553591002e-05,
+      "loss": 0.0446,
+      "step": 990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2947866202164234e-05,
+      "loss": 0.0331,
+      "step": 991
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2934656219438657e-05,
+      "loss": 0.0388,
+      "step": 992
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2921440630649257e-05,
+      "loss": 0.0356,
+      "step": 993
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2908219461041724e-05,
+      "loss": 0.0331,
+      "step": 994
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2894992735872417e-05,
+      "loss": 0.0364,
+      "step": 995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.28817604804083e-05,
+      "loss": 0.0433,
+      "step": 996
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2868522719926913e-05,
+      "loss": 0.0511,
+      "step": 997
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2855279479716298e-05,
+      "loss": 0.0482,
+      "step": 998
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2842030785074976e-05,
+      "loss": 0.0446,
+      "step": 999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2828776661311884e-05,
+      "loss": 0.0434,
+      "step": 1000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2815517133746324e-05,
+      "loss": 0.0399,
+      "step": 1001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2802252227707931e-05,
+      "loss": 0.0492,
+      "step": 1002
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2788981968536614e-05,
+      "loss": 0.033,
+      "step": 1003
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2775706381582493e-05,
+      "loss": 0.0395,
+      "step": 1004
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2762425492205887e-05,
+      "loss": 0.0455,
+      "step": 1005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.274913932577723e-05,
+      "loss": 0.0603,
+      "step": 1006
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2735847907677036e-05,
+      "loss": 0.0361,
+      "step": 1007
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2722551263295865e-05,
+      "loss": 0.0496,
+      "step": 1008
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2709249418034249e-05,
+      "loss": 0.0343,
+      "step": 1009
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.2695942397302654e-05,
+      "loss": 0.0355,
+      "step": 1010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.268263022652144e-05,
+      "loss": 0.0301,
+      "step": 1011
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.26693129311208e-05,
+      "loss": 0.0466,
+      "step": 1012
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.265599053654072e-05,
+      "loss": 0.0501,
+      "step": 1013
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.264266306823092e-05,
+      "loss": 0.0388,
+      "step": 1014
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2629330551650827e-05,
+      "loss": 0.0532,
+      "step": 1015
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2615993012269492e-05,
+      "loss": 0.0289,
+      "step": 1016
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2602650475565578e-05,
+      "loss": 0.0312,
+      "step": 1017
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2589302967027286e-05,
+      "loss": 0.0427,
+      "step": 1018
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2575950512152315e-05,
+      "loss": 0.0414,
+      "step": 1019
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2562593136447814e-05,
+      "loss": 0.0333,
+      "step": 1020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2549230865430337e-05,
+      "loss": 0.0454,
+      "step": 1021
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.253586372462578e-05,
+      "loss": 0.0455,
+      "step": 1022
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.252249173956935e-05,
+      "loss": 0.0431,
+      "step": 1023
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.2509114935805502e-05,
+      "loss": 0.0423,
+      "step": 1024
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.24957333388879e-05,
+      "loss": 0.0442,
+      "step": 1025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2482346974379365e-05,
+      "loss": 0.0484,
+      "step": 1026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2468955867851817e-05,
+      "loss": 0.0569,
+      "step": 1027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.245556004488625e-05,
+      "loss": 0.046,
+      "step": 1028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2442159531072657e-05,
+      "loss": 0.0448,
+      "step": 1029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2428754352009989e-05,
+      "loss": 0.0405,
+      "step": 1030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2415344533306121e-05,
+      "loss": 0.0429,
+      "step": 1031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.240193010057778e-05,
+      "loss": 0.0424,
+      "step": 1032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2388511079450512e-05,
+      "loss": 0.0418,
+      "step": 1033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2375087495558629e-05,
+      "loss": 0.0516,
+      "step": 1034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2361659374545157e-05,
+      "loss": 0.0373,
+      "step": 1035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2348226742061791e-05,
+      "loss": 0.0453,
+      "step": 1036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.2334789623768838e-05,
+      "loss": 0.0375,
+      "step": 1037
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2321348045335183e-05,
+      "loss": 0.029,
+      "step": 1038
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2307902032438233e-05,
+      "loss": 0.045,
+      "step": 1039
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2294451610763852e-05,
+      "loss": 0.0387,
+      "step": 1040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2280996806006343e-05,
+      "loss": 0.0383,
+      "step": 1041
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2267537643868366e-05,
+      "loss": 0.0334,
+      "step": 1042
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2254074150060917e-05,
+      "loss": 0.0452,
+      "step": 1043
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2240606350303266e-05,
+      "loss": 0.0467,
+      "step": 1044
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.22271342703229e-05,
+      "loss": 0.0467,
+      "step": 1045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2213657935855487e-05,
+      "loss": 0.0309,
+      "step": 1046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2200177372644828e-05,
+      "loss": 0.0301,
+      "step": 1047
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.2186692606442794e-05,
+      "loss": 0.0361,
+      "step": 1048
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2173203663009287e-05,
+      "loss": 0.0424,
+      "step": 1049
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2159710568112191e-05,
+      "loss": 0.0399,
+      "step": 1050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.214621334752732e-05,
+      "loss": 0.0495,
+      "step": 1051
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2132712027038372e-05,
+      "loss": 0.0441,
+      "step": 1052
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2119206632436866e-05,
+      "loss": 0.0454,
+      "step": 1053
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2105697189522121e-05,
+      "loss": 0.0402,
+      "step": 1054
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2092183724101171e-05,
+      "loss": 0.0436,
+      "step": 1055
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2078666261988748e-05,
+      "loss": 0.0283,
+      "step": 1056
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2065144829007213e-05,
+      "loss": 0.0461,
+      "step": 1057
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2051619450986516e-05,
+      "loss": 0.0368,
+      "step": 1058
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2038090153764134e-05,
+      "loss": 0.0393,
+      "step": 1059
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2024556963185045e-05,
+      "loss": 0.0427,
+      "step": 1060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.2011019905101653e-05,
+      "loss": 0.0337,
+      "step": 1061
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1997479005373757e-05,
+      "loss": 0.0356,
+      "step": 1062
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1983934289868488e-05,
+      "loss": 0.0397,
+      "step": 1063
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1970385784460274e-05,
+      "loss": 0.0403,
+      "step": 1064
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1956833515030776e-05,
+      "loss": 0.0291,
+      "step": 1065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1943277507468848e-05,
+      "loss": 0.0448,
+      "step": 1066
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1929717787670492e-05,
+      "loss": 0.0437,
+      "step": 1067
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1916154381538787e-05,
+      "loss": 0.0434,
+      "step": 1068
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.190258731498386e-05,
+      "loss": 0.047,
+      "step": 1069
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.188901661392284e-05,
+      "loss": 0.0629,
+      "step": 1070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.187544230427979e-05,
+      "loss": 0.0455,
+      "step": 1071
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1861864411985662e-05,
+      "loss": 0.0409,
+      "step": 1072
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1848282962978262e-05,
+      "loss": 0.0401,
+      "step": 1073
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1834697983202186e-05,
+      "loss": 0.0439,
+      "step": 1074
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1821109498608774e-05,
+      "loss": 0.0336,
+      "step": 1075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.180751753515606e-05,
+      "loss": 0.039,
+      "step": 1076
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1793922118808727e-05,
+      "loss": 0.0454,
+      "step": 1077
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1780323275538057e-05,
+      "loss": 0.0388,
+      "step": 1078
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1766721031321867e-05,
+      "loss": 0.0347,
+      "step": 1079
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1753115412144482e-05,
+      "loss": 0.0325,
+      "step": 1080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1739506443996673e-05,
+      "loss": 0.0446,
+      "step": 1081
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1725894152875599e-05,
+      "loss": 0.0495,
+      "step": 1082
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.1712278564784775e-05,
+      "loss": 0.0327,
+      "step": 1083
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.169865970573402e-05,
+      "loss": 0.0487,
+      "step": 1084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1685037601739385e-05,
+      "loss": 0.0305,
+      "step": 1085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1671412278823134e-05,
+      "loss": 0.0594,
+      "step": 1086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1657783763013675e-05,
+      "loss": 0.0321,
+      "step": 1087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1644152080345517e-05,
+      "loss": 0.0309,
+      "step": 1088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.163051725685921e-05,
+      "loss": 0.0342,
+      "step": 1089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1616879318601319e-05,
+      "loss": 0.0367,
+      "step": 1090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1603238291624348e-05,
+      "loss": 0.0334,
+      "step": 1091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1589594201986704e-05,
+      "loss": 0.0281,
+      "step": 1092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1575947075752646e-05,
+      "loss": 0.0287,
+      "step": 1093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1562296938992233e-05,
+      "loss": 0.0277,
+      "step": 1094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1548643817781273e-05,
+      "loss": 0.0336,
+      "step": 1095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1534987738201284e-05,
+      "loss": 0.037,
+      "step": 1096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.152132872633942e-05,
+      "loss": 0.027,
+      "step": 1097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1507666808288451e-05,
+      "loss": 0.0546,
+      "step": 1098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1494002010146688e-05,
+      "loss": 0.0365,
+      "step": 1099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1480334358017952e-05,
+      "loss": 0.0372,
+      "step": 1100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1466663878011512e-05,
+      "loss": 0.0315,
+      "step": 1101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1452990596242038e-05,
+      "loss": 0.0462,
+      "step": 1102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1439314538829556e-05,
+      "loss": 0.0427,
+      "step": 1103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.142563573189939e-05,
+      "loss": 0.0534,
+      "step": 1104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1411954201582116e-05,
+      "loss": 0.0374,
+      "step": 1105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1398269974013522e-05,
+      "loss": 0.0419,
+      "step": 1106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.138458307533454e-05,
+      "loss": 0.0378,
+      "step": 1107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1370893531691201e-05,
+      "loss": 0.0483,
+      "step": 1108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1357201369234599e-05,
+      "loss": 0.0369,
+      "step": 1109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1343506614120822e-05,
+      "loss": 0.0452,
+      "step": 1110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1329809292510916e-05,
+      "loss": 0.0466,
+      "step": 1111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1316109430570828e-05,
+      "loss": 0.0635,
+      "step": 1112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1302407054471355e-05,
+      "loss": 0.0452,
+      "step": 1113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1288702190388103e-05,
+      "loss": 0.0414,
+      "step": 1114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.127499486450143e-05,
+      "loss": 0.0459,
+      "step": 1115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1261285102996386e-05,
+      "loss": 0.0384,
+      "step": 1116
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1247572932062689e-05,
+      "loss": 0.0418,
+      "step": 1117
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1233858377894648e-05,
+      "loss": 0.0445,
+      "step": 1118
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1220141466691134e-05,
+      "loss": 0.0446,
+      "step": 1119
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1206422224655515e-05,
+      "loss": 0.0404,
+      "step": 1120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1192700677995607e-05,
+      "loss": 0.0493,
+      "step": 1121
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.117897685292364e-05,
+      "loss": 0.0417,
+      "step": 1122
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1165250775656189e-05,
+      "loss": 0.0382,
+      "step": 1123
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1151522472414134e-05,
+      "loss": 0.0325,
+      "step": 1124
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.11377919694226e-05,
+      "loss": 0.0354,
+      "step": 1125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1124059292910926e-05,
+      "loss": 0.041,
+      "step": 1126
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.11103244691126e-05,
+      "loss": 0.0451,
+      "step": 1127
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1096587524265199e-05,
+      "loss": 0.0469,
+      "step": 1128
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1082848484610373e-05,
+      "loss": 0.0409,
+      "step": 1129
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.106910737639376e-05,
+      "loss": 0.0521,
+      "step": 1130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1055364225864948e-05,
+      "loss": 0.0306,
+      "step": 1131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1041619059277436e-05,
+      "loss": 0.0467,
+      "step": 1132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1027871902888567e-05,
+      "loss": 0.045,
+      "step": 1133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1014122782959486e-05,
+      "loss": 0.0376,
+      "step": 1134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.1000371725755093e-05,
+      "loss": 0.0415,
+      "step": 1135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0986618757543982e-05,
+      "loss": 0.0323,
+      "step": 1136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0972863904598406e-05,
+      "loss": 0.0407,
+      "step": 1137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0959107193194208e-05,
+      "loss": 0.044,
+      "step": 1138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0945348649610794e-05,
+      "loss": 0.0332,
+      "step": 1139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0931588300131062e-05,
+      "loss": 0.0413,
+      "step": 1140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0917826171041355e-05,
+      "loss": 0.0515,
+      "step": 1141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0904062288631429e-05,
+      "loss": 0.0349,
+      "step": 1142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0890296679194379e-05,
+      "loss": 0.0281,
+      "step": 1143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0876529369026598e-05,
+      "loss": 0.0402,
+      "step": 1144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0862760384427737e-05,
+      "loss": 0.0383,
+      "step": 1145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0848989751700643e-05,
+      "loss": 0.0517,
+      "step": 1146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0835217497151298e-05,
+      "loss": 0.0318,
+      "step": 1147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0821443647088803e-05,
+      "loss": 0.044,
+      "step": 1148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0807668227825297e-05,
+      "loss": 0.0379,
+      "step": 1149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0793891265675913e-05,
+      "loss": 0.0402,
+      "step": 1150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0780112786958734e-05,
+      "loss": 0.0439,
+      "step": 1151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0766332817994745e-05,
+      "loss": 0.048,
+      "step": 1152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0752551385107772e-05,
+      "loss": 0.0327,
+      "step": 1153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.073876851462444e-05,
+      "loss": 0.0397,
+      "step": 1154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.072498423287412e-05,
+      "loss": 0.0573,
+      "step": 1155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0711198566188878e-05,
+      "loss": 0.0494,
+      "step": 1156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0697411540903424e-05,
+      "loss": 0.0407,
+      "step": 1157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0683623183355072e-05,
+      "loss": 0.0343,
+      "step": 1158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0669833519883667e-05,
+      "loss": 0.0323,
+      "step": 1159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0656042576831564e-05,
+      "loss": 0.0531,
+      "step": 1160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.064225038054355e-05,
+      "loss": 0.0384,
+      "step": 1161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0628456957366811e-05,
+      "loss": 0.0453,
+      "step": 1162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0614662333650879e-05,
+      "loss": 0.0375,
+      "step": 1163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.060086653574757e-05,
+      "loss": 0.0353,
+      "step": 1164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0587069590010961e-05,
+      "loss": 0.0432,
+      "step": 1165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0573271522797304e-05,
+      "loss": 0.0466,
+      "step": 1166
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0559472360464997e-05,
+      "loss": 0.0455,
+      "step": 1167
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0545672129374542e-05,
+      "loss": 0.0514,
+      "step": 1168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0531870855888464e-05,
+      "loss": 0.0348,
+      "step": 1169
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0518068566371295e-05,
+      "loss": 0.0336,
+      "step": 1170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0504265287189502e-05,
+      "loss": 0.0329,
+      "step": 1171
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.049046104471144e-05,
+      "loss": 0.0345,
+      "step": 1172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0476655865307308e-05,
+      "loss": 0.0392,
+      "step": 1173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0462849775349091e-05,
+      "loss": 0.0452,
+      "step": 1174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0449042801210519e-05,
+      "loss": 0.033,
+      "step": 1175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0435234969267006e-05,
+      "loss": 0.0311,
+      "step": 1176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0421426305895606e-05,
+      "loss": 0.0305,
+      "step": 1177
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0407616837474964e-05,
+      "loss": 0.0343,
+      "step": 1178
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0393806590385261e-05,
+      "loss": 0.0322,
+      "step": 1179
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0379995591008159e-05,
+      "loss": 0.0285,
+      "step": 1180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.036618386572677e-05,
+      "loss": 0.0416,
+      "step": 1181
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0352371440925586e-05,
+      "loss": 0.0437,
+      "step": 1182
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0338558342990432e-05,
+      "loss": 0.0408,
+      "step": 1183
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0324744598308422e-05,
+      "loss": 0.0399,
+      "step": 1184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.031093023326791e-05,
+      "loss": 0.0393,
+      "step": 1185
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.0297115274258427e-05,
+      "loss": 0.0339,
+      "step": 1186
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.028329974767064e-05,
+      "loss": 0.0252,
+      "step": 1187
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.026948367989631e-05,
+      "loss": 0.0302,
+      "step": 1188
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.025566709732822e-05,
+      "loss": 0.0344,
+      "step": 1189
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0241850026360135e-05,
+      "loss": 0.0409,
+      "step": 1190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.022803249338677e-05,
+      "loss": 0.045,
+      "step": 1191
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0214214524803702e-05,
+      "loss": 0.0383,
+      "step": 1192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0200396147007356e-05,
+      "loss": 0.0307,
+      "step": 1193
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0186577386394932e-05,
+      "loss": 0.0307,
+      "step": 1194
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0172758269364358e-05,
+      "loss": 0.0445,
+      "step": 1195
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0158938822314248e-05,
+      "loss": 0.0433,
+      "step": 1196
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0145119071643843e-05,
+      "loss": 0.0306,
+      "step": 1197
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.013129904375297e-05,
+      "loss": 0.0296,
+      "step": 1198
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0117478765041976e-05,
+      "loss": 0.0284,
+      "step": 1199
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0103658261911698e-05,
+      "loss": 0.0327,
+      "step": 1200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.0089837560763393e-05,
+      "loss": 0.0314,
+      "step": 1201
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.00760166879987e-05,
+      "loss": 0.0327,
+      "step": 1202
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0062195670019583e-05,
+      "loss": 0.0287,
+      "step": 1203
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0048374533228293e-05,
+      "loss": 0.0381,
+      "step": 1204
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0034553304027294e-05,
+      "loss": 0.0379,
+      "step": 1205
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0020732008819236e-05,
+      "loss": 0.0364,
+      "step": 1206
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.0006910674006891e-05,
+      "loss": 0.0338,
+      "step": 1207
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.993089325993112e-06,
+      "loss": 0.0418,
+      "step": 1208
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.979267991180768e-06,
+      "loss": 0.0339,
+      "step": 1209
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.96544669597271e-06,
+      "loss": 0.058,
+      "step": 1210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.951625466771708e-06,
+      "loss": 0.0436,
+      "step": 1211
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.937804329980417e-06,
+      "loss": 0.041,
+      "step": 1212
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.923983312001303e-06,
+      "loss": 0.0352,
+      "step": 1213
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.910162439236609e-06,
+      "loss": 0.0468,
+      "step": 1214
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.896341738088306e-06,
+      "loss": 0.0331,
+      "step": 1215
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.882521234958027e-06,
+      "loss": 0.0328,
+      "step": 1216
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.868700956247035e-06,
+      "loss": 0.0276,
+      "step": 1217
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.854880928356157e-06,
+      "loss": 0.0469,
+      "step": 1218
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.841061177685755e-06,
+      "loss": 0.0297,
+      "step": 1219
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.827241730635645e-06,
+      "loss": 0.0313,
+      "step": 1220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.813422613605072e-06,
+      "loss": 0.0311,
+      "step": 1221
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.799603852992646e-06,
+      "loss": 0.0434,
+      "step": 1222
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.7857854751963e-06,
+      "loss": 0.031,
+      "step": 1223
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.771967506613232e-06,
+      "loss": 0.0454,
+      "step": 1224
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.758149973639866e-06,
+      "loss": 0.0285,
+      "step": 1225
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.744332902671784e-06,
+      "loss": 0.0264,
+      "step": 1226
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.730516320103693e-06,
+      "loss": 0.0304,
+      "step": 1227
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.716700252329361e-06,
+      "loss": 0.0319,
+      "step": 1228
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.702884725741578e-06,
+      "loss": 0.0279,
+      "step": 1229
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.689069766732095e-06,
+      "loss": 0.0378,
+      "step": 1230
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.675255401691578e-06,
+      "loss": 0.0188,
+      "step": 1231
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.66144165700957e-06,
+      "loss": 0.0318,
+      "step": 1232
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.647628559074417e-06,
+      "loss": 0.0565,
+      "step": 1233
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.633816134273233e-06,
+      "loss": 0.0403,
+      "step": 1234
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.620004408991843e-06,
+      "loss": 0.0283,
+      "step": 1235
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.606193409614744e-06,
+      "loss": 0.0235,
+      "step": 1236
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.592383162525038e-06,
+      "loss": 0.0297,
+      "step": 1237
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.578573694104396e-06,
+      "loss": 0.0406,
+      "step": 1238
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.564765030732998e-06,
+      "loss": 0.043,
+      "step": 1239
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.550957198789484e-06,
+      "loss": 0.0343,
+      "step": 1240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.537150224650912e-06,
+      "loss": 0.0435,
+      "step": 1241
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.523344134692699e-06,
+      "loss": 0.042,
+      "step": 1242
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.509538955288566e-06,
+      "loss": 0.0296,
+      "step": 1243
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.4957347128105e-06,
+      "loss": 0.0494,
+      "step": 1244
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.481931433628707e-06,
+      "loss": 0.0293,
+      "step": 1245
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.468129144111539e-06,
+      "loss": 0.0346,
+      "step": 1246
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.454327870625462e-06,
+      "loss": 0.0355,
+      "step": 1247
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.440527639535006e-06,
+      "loss": 0.0303,
+      "step": 1248
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.426728477202702e-06,
+      "loss": 0.0307,
+      "step": 1249
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.41293040998904e-06,
+      "loss": 0.0371,
+      "step": 1250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.39913346425243e-06,
+      "loss": 0.0451,
+      "step": 1251
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.385337666349126e-06,
+      "loss": 0.032,
+      "step": 1252
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.371543042633194e-06,
+      "loss": 0.038,
+      "step": 1253
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.357749619456456e-06,
+      "loss": 0.0359,
+      "step": 1254
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.343957423168441e-06,
+      "loss": 0.0338,
+      "step": 1255
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.330166480116334e-06,
+      "loss": 0.0374,
+      "step": 1256
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.316376816644931e-06,
+      "loss": 0.0312,
+      "step": 1257
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.302588459096576e-06,
+      "loss": 0.0411,
+      "step": 1258
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.288801433811125e-06,
+      "loss": 0.0347,
+      "step": 1259
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.275015767125883e-06,
+      "loss": 0.0268,
+      "step": 1260
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.261231485375564e-06,
+      "loss": 0.0374,
+      "step": 1261
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.247448614892232e-06,
+      "loss": 0.0412,
+      "step": 1262
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.23366718200526e-06,
+      "loss": 0.0404,
+      "step": 1263
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.219887213041267e-06,
+      "loss": 0.0342,
+      "step": 1264
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.20610873432409e-06,
+      "loss": 0.0276,
+      "step": 1265
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.192331772174706e-06,
+      "loss": 0.034,
+      "step": 1266
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.1785563529112e-06,
+      "loss": 0.032,
+      "step": 1267
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.164782502848703e-06,
+      "loss": 0.0338,
+      "step": 1268
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.151010248299364e-06,
+      "loss": 0.0557,
+      "step": 1269
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.137239615572263e-06,
+      "loss": 0.0333,
+      "step": 1270
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.123470630973404e-06,
+      "loss": 0.0283,
+      "step": 1271
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.109703320805624e-06,
+      "loss": 0.0414,
+      "step": 1272
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.095937711368574e-06,
+      "loss": 0.0466,
+      "step": 1273
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.082173828958647e-06,
+      "loss": 0.0393,
+      "step": 1274
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.068411699868944e-06,
+      "loss": 0.0232,
+      "step": 1275
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.054651350389211e-06,
+      "loss": 0.0358,
+      "step": 1276
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.040892806805794e-06,
+      "loss": 0.0455,
+      "step": 1277
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.0271360954016e-06,
+      "loss": 0.0551,
+      "step": 1278
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.01338124245602e-06,
+      "loss": 0.0261,
+      "step": 1279
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.999628274244912e-06,
+      "loss": 0.0347,
+      "step": 1280
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.98587721704052e-06,
+      "loss": 0.036,
+      "step": 1281
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.97212809711144e-06,
+      "loss": 0.0422,
+      "step": 1282
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.958380940722566e-06,
+      "loss": 0.0501,
+      "step": 1283
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.944635774135054e-06,
+      "loss": 0.0345,
+      "step": 1284
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.930892623606243e-06,
+      "loss": 0.0329,
+      "step": 1285
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.917151515389629e-06,
+      "loss": 0.0446,
+      "step": 1286
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.903412475734804e-06,
+      "loss": 0.0312,
+      "step": 1287
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.889675530887406e-06,
+      "loss": 0.029,
+      "step": 1288
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.875940707089076e-06,
+      "loss": 0.0375,
+      "step": 1289
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.862208030577403e-06,
+      "loss": 0.0396,
+      "step": 1290
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.848477527585871e-06,
+      "loss": 0.0366,
+      "step": 1291
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.834749224343814e-06,
+      "loss": 0.0389,
+      "step": 1292
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.821023147076364e-06,
+      "loss": 0.0347,
+      "step": 1293
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.807299322004396e-06,
+      "loss": 0.0379,
+      "step": 1294
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.79357777534449e-06,
+      "loss": 0.0294,
+      "step": 1295
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.779858533308866e-06,
+      "loss": 0.0264,
+      "step": 1296
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.766141622105352e-06,
+      "loss": 0.0348,
+      "step": 1297
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.752427067937314e-06,
+      "loss": 0.0415,
+      "step": 1298
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.738714897003617e-06,
+      "loss": 0.0296,
+      "step": 1299
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.725005135498574e-06,
+      "loss": 0.0326,
+      "step": 1300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.711297809611898e-06,
+      "loss": 0.0363,
+      "step": 1301
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.697592945528648e-06,
+      "loss": 0.0397,
+      "step": 1302
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.683890569429174e-06,
+      "loss": 0.0308,
+      "step": 1303
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.670190707489085e-06,
+      "loss": 0.0341,
+      "step": 1304
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.656493385879181e-06,
+      "loss": 0.0395,
+      "step": 1305
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.642798630765404e-06,
+      "loss": 0.0322,
+      "step": 1306
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.629106468308802e-06,
+      "loss": 0.0479,
+      "step": 1307
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.615416924665464e-06,
+      "loss": 0.0234,
+      "step": 1308
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.601730025986476e-06,
+      "loss": 0.0367,
+      "step": 1309
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.588045798417884e-06,
+      "loss": 0.0351,
+      "step": 1310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.574364268100614e-06,
+      "loss": 0.0423,
+      "step": 1311
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.560685461170447e-06,
+      "loss": 0.0307,
+      "step": 1312
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.547009403757964e-06,
+      "loss": 0.0506,
+      "step": 1313
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.533336121988491e-06,
+      "loss": 0.0366,
+      "step": 1314
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.519665641982053e-06,
+      "loss": 0.0337,
+      "step": 1315
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.505997989853312e-06,
+      "loss": 0.0296,
+      "step": 1316
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.49233319171155e-06,
+      "loss": 0.0267,
+      "step": 1317
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.478671273660582e-06,
+      "loss": 0.0379,
+      "step": 1318
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.46501226179872e-06,
+      "loss": 0.0274,
+      "step": 1319
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.451356182218728e-06,
+      "loss": 0.044,
+      "step": 1320
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.437703061007769e-06,
+      "loss": 0.0382,
+      "step": 1321
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.424052924247356e-06,
+      "loss": 0.0424,
+      "step": 1322
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.4104057980133e-06,
+      "loss": 0.029,
+      "step": 1323
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.396761708375653e-06,
+      "loss": 0.0459,
+      "step": 1324
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.383120681398683e-06,
+      "loss": 0.0371,
+      "step": 1325
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.369482743140793e-06,
+      "loss": 0.0305,
+      "step": 1326
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.35584791965449e-06,
+      "loss": 0.0295,
+      "step": 1327
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.342216236986329e-06,
+      "loss": 0.0305,
+      "step": 1328
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.328587721176866e-06,
+      "loss": 0.0277,
+      "step": 1329
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.314962398260618e-06,
+      "loss": 0.0309,
+      "step": 1330
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.301340294265986e-06,
+      "loss": 0.0235,
+      "step": 1331
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.287721435215228e-06,
+      "loss": 0.0459,
+      "step": 1332
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.274105847124404e-06,
+      "loss": 0.0296,
+      "step": 1333
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.260493556003332e-06,
+      "loss": 0.0257,
+      "step": 1334
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.24688458785552e-06,
+      "loss": 0.026,
+      "step": 1335
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.233278968678135e-06,
+      "loss": 0.0396,
+      "step": 1336
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.219676724461946e-06,
+      "loss": 0.0322,
+      "step": 1337
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.206077881191275e-06,
+      "loss": 0.0323,
+      "step": 1338
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.192482464843944e-06,
+      "loss": 0.0448,
+      "step": 1339
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.178890501391233e-06,
+      "loss": 0.0366,
+      "step": 1340
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.16530201679782e-06,
+      "loss": 0.0356,
+      "step": 1341
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.151717037021738e-06,
+      "loss": 0.0256,
+      "step": 1342
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.13813558801434e-06,
+      "loss": 0.027,
+      "step": 1343
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.124557695720214e-06,
+      "loss": 0.0274,
+      "step": 1344
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.110983386077161e-06,
+      "loss": 0.0413,
+      "step": 1345
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.097412685016142e-06,
+      "loss": 0.0414,
+      "step": 1346
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.083845618461218e-06,
+      "loss": 0.0371,
+      "step": 1347
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.07028221232951e-06,
+      "loss": 0.0434,
+      "step": 1348
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.056722492531152e-06,
+      "loss": 0.03,
+      "step": 1349
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.043166484969225e-06,
+      "loss": 0.0257,
+      "step": 1350
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.029614215539729e-06,
+      "loss": 0.0283,
+      "step": 1351
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.016065710131515e-06,
+      "loss": 0.0345,
+      "step": 1352
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.002520994626248e-06,
+      "loss": 0.0407,
+      "step": 1353
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.98898009489835e-06,
+      "loss": 0.0332,
+      "step": 1354
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.975443036814956e-06,
+      "loss": 0.0457,
+      "step": 1355
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.961909846235868e-06,
+      "loss": 0.0339,
+      "step": 1356
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.948380549013487e-06,
+      "loss": 0.0409,
+      "step": 1357
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.934855170992789e-06,
+      "loss": 0.0335,
+      "step": 1358
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.921333738011255e-06,
+      "loss": 0.0325,
+      "step": 1359
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.907816275898834e-06,
+      "loss": 0.0315,
+      "step": 1360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.89430281047788e-06,
+      "loss": 0.0242,
+      "step": 1361
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.880793367563134e-06,
+      "loss": 0.034,
+      "step": 1362
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.86728797296163e-06,
+      "loss": 0.0297,
+      "step": 1363
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.853786652472682e-06,
+      "loss": 0.0401,
+      "step": 1364
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 7.840289431887814e-06,
+      "loss": 0.0187,
+      "step": 1365
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.826796336990716e-06,
+      "loss": 0.0426,
+      "step": 1366
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.81330739355721e-06,
+      "loss": 0.0432,
+      "step": 1367
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.799822627355172e-06,
+      "loss": 0.0429,
+      "step": 1368
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.786342064144513e-06,
+      "loss": 0.0387,
+      "step": 1369
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.772865729677102e-06,
+      "loss": 0.0284,
+      "step": 1370
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.759393649696737e-06,
+      "loss": 0.0298,
+      "step": 1371
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.745925849939085e-06,
+      "loss": 0.0449,
+      "step": 1372
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.732462356131639e-06,
+      "loss": 0.0385,
+      "step": 1373
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.719003193993664e-06,
+      "loss": 0.0463,
+      "step": 1374
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.70554838923615e-06,
+      "loss": 0.0285,
+      "step": 1375
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.692097967561772e-06,
+      "loss": 0.0394,
+      "step": 1376
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 7.67865195466482e-06,
+      "loss": 0.0435,
+      "step": 1377
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.665210376231166e-06,
+      "loss": 0.0332,
+      "step": 1378
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.651773257938215e-06,
+      "loss": 0.0394,
+      "step": 1379
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.638340625454848e-06,
+      "loss": 0.0464,
+      "step": 1380
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.624912504441371e-06,
+      "loss": 0.0387,
+      "step": 1381
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.6114889205494876e-06,
+      "loss": 0.0401,
+      "step": 1382
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5980698994222225e-06,
+      "loss": 0.041,
+      "step": 1383
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.584655466693882e-06,
+      "loss": 0.0299,
+      "step": 1384
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.5712456479900135e-06,
+      "loss": 0.0288,
+      "step": 1385
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.557840468927349e-06,
+      "loss": 0.0338,
+      "step": 1386
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.544439955113753e-06,
+      "loss": 0.0264,
+      "step": 1387
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 7.531044132148184e-06,
+      "loss": 0.0289,
+      "step": 1388
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.517653025620638e-06,
+      "loss": 0.0346,
+      "step": 1389
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.504266661112103e-06,
+      "loss": 0.0415,
+      "step": 1390
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.490885064194502e-06,
+      "loss": 0.044,
+      "step": 1391
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.477508260430655e-06,
+      "loss": 0.0403,
+      "step": 1392
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.464136275374224e-06,
+      "loss": 0.0438,
+      "step": 1393
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.450769134569664e-06,
+      "loss": 0.035,
+      "step": 1394
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.437406863552186e-06,
+      "loss": 0.0346,
+      "step": 1395
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.424049487847688e-06,
+      "loss": 0.0304,
+      "step": 1396
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.410697032972718e-06,
+      "loss": 0.0316,
+      "step": 1397
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.397349524434425e-06,
+      "loss": 0.0312,
+      "step": 1398
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.384006987730512e-06,
+      "loss": 0.0333,
+      "step": 1399
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 7.370669448349179e-06,
+      "loss": 0.0325,
+      "step": 1400
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.35733693176908e-06,
+      "loss": 0.0411,
+      "step": 1401
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.344009463459283e-06,
+      "loss": 0.043,
+      "step": 1402
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.330687068879203e-06,
+      "loss": 0.0258,
+      "step": 1403
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.317369773478564e-06,
+      "loss": 0.0373,
+      "step": 1404
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.30405760269735e-06,
+      "loss": 0.045,
+      "step": 1405
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.2907505819657555e-06,
+      "loss": 0.0346,
+      "step": 1406
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.277448736704133e-06,
+      "loss": 0.0416,
+      "step": 1407
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.264152092322964e-06,
+      "loss": 0.0274,
+      "step": 1408
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.250860674222775e-06,
+      "loss": 0.0443,
+      "step": 1409
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.237574507794115e-06,
+      "loss": 0.0327,
+      "step": 1410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.224293618417508e-06,
+      "loss": 0.0276,
+      "step": 1411
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 7.211018031463391e-06,
+      "loss": 0.0471,
+      "step": 1412
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.197747772292072e-06,
+      "loss": 0.0401,
+      "step": 1413
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.184482866253677e-06,
+      "loss": 0.0282,
+      "step": 1414
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.171223338688118e-06,
+      "loss": 0.0276,
+      "step": 1415
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.157969214925025e-06,
+      "loss": 0.0359,
+      "step": 1416
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.144720520283703e-06,
+      "loss": 0.0339,
+      "step": 1417
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.131477280073091e-06,
+      "loss": 0.0387,
+      "step": 1418
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.1182395195917006e-06,
+      "loss": 0.0339,
+      "step": 1419
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.105007264127586e-06,
+      "loss": 0.0336,
+      "step": 1420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.09178053895828e-06,
+      "loss": 0.0388,
+      "step": 1421
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.078559369350746e-06,
+      "loss": 0.0378,
+      "step": 1422
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 7.065343780561345e-06,
+      "loss": 0.0418,
+      "step": 1423
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.0521337978357686e-06,
+      "loss": 0.0283,
+      "step": 1424
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.038929446409004e-06,
+      "loss": 0.0359,
+      "step": 1425
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.025730751505275e-06,
+      "loss": 0.0303,
+      "step": 1426
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.012537738338005e-06,
+      "loss": 0.0217,
+      "step": 1427
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.999350432109766e-06,
+      "loss": 0.0366,
+      "step": 1428
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.986168858012222e-06,
+      "loss": 0.0334,
+      "step": 1429
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9729930412260905e-06,
+      "loss": 0.0282,
+      "step": 1430
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.9598230069210845e-06,
+      "loss": 0.0238,
+      "step": 1431
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.946658780255882e-06,
+      "loss": 0.0324,
+      "step": 1432
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.933500386378057e-06,
+      "loss": 0.0311,
+      "step": 1433
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.920347850424045e-06,
+      "loss": 0.0296,
+      "step": 1434
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 6.907201197519095e-06,
+      "loss": 0.0394,
+      "step": 1435
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.894060452777209e-06,
+      "loss": 0.026,
+      "step": 1436
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.880925641301109e-06,
+      "loss": 0.0332,
+      "step": 1437
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.867796788182182e-06,
+      "loss": 0.0371,
+      "step": 1438
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.854673918500427e-06,
+      "loss": 0.0415,
+      "step": 1439
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.841557057324417e-06,
+      "loss": 0.0385,
+      "step": 1440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.828446229711255e-06,
+      "loss": 0.0419,
+      "step": 1441
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.815341460706506e-06,
+      "loss": 0.0287,
+      "step": 1442
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.8022427753441644e-06,
+      "loss": 0.025,
+      "step": 1443
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.789150198646608e-06,
+      "loss": 0.0356,
+      "step": 1444
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.77606375562454e-06,
+      "loss": 0.0353,
+      "step": 1445
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.762983471276943e-06,
+      "loss": 0.0342,
+      "step": 1446
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 6.749909370591049e-06,
+      "loss": 0.034,
+      "step": 1447
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.736841478542264e-06,
+      "loss": 0.045,
+      "step": 1448
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.723779820094138e-06,
+      "loss": 0.0272,
+      "step": 1449
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.710724420198312e-06,
+      "loss": 0.0204,
+      "step": 1450
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.697675303794473e-06,
+      "loss": 0.0349,
+      "step": 1451
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6846324958103015e-06,
+      "loss": 0.0338,
+      "step": 1452
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.671596021161431e-06,
+      "loss": 0.0375,
+      "step": 1453
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.6585659047514e-06,
+      "loss": 0.0374,
+      "step": 1454
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.645542171471586e-06,
+      "loss": 0.0375,
+      "step": 1455
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.632524846201189e-06,
+      "loss": 0.0373,
+      "step": 1456
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.619513953807161e-06,
+      "loss": 0.0332,
+      "step": 1457
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.606509519144166e-06,
+      "loss": 0.041,
+      "step": 1458
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 6.593511567054523e-06,
+      "loss": 0.036,
+      "step": 1459
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.580520122368189e-06,
+      "loss": 0.024,
+      "step": 1460
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.567535209902669e-06,
+      "loss": 0.0299,
+      "step": 1461
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.554556854462999e-06,
+      "loss": 0.0385,
+      "step": 1462
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.5415850808416884e-06,
+      "loss": 0.0365,
+      "step": 1463
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.528619913818668e-06,
+      "loss": 0.0407,
+      "step": 1464
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.515661378161257e-06,
+      "loss": 0.033,
+      "step": 1465
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.502709498624098e-06,
+      "loss": 0.0361,
+      "step": 1466
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.489764299949126e-06,
+      "loss": 0.0358,
+      "step": 1467
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.476825806865511e-06,
+      "loss": 0.0299,
+      "step": 1468
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.46389404408961e-06,
+      "loss": 0.0354,
+      "step": 1469
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 6.450969036324927e-06,
+      "loss": 0.0339,
+      "step": 1470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.43805080826206e-06,
+      "loss": 0.0327,
+      "step": 1471
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.425139384578656e-06,
+      "loss": 0.0329,
+      "step": 1472
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.41223478993936e-06,
+      "loss": 0.0333,
+      "step": 1473
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3993370489957825e-06,
+      "loss": 0.0287,
+      "step": 1474
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3864461863864325e-06,
+      "loss": 0.0511,
+      "step": 1475
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.373562226736676e-06,
+      "loss": 0.0275,
+      "step": 1476
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.360685194658705e-06,
+      "loss": 0.0367,
+      "step": 1477
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.347815114751466e-06,
+      "loss": 0.0417,
+      "step": 1478
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.3349520116006234e-06,
+      "loss": 0.023,
+      "step": 1479
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.322095909778532e-06,
+      "loss": 0.0276,
+      "step": 1480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.309246833844152e-06,
+      "loss": 0.0374,
+      "step": 1481
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 6.296404808343031e-06,
+      "loss": 0.0466,
+      "step": 1482
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.283569857807246e-06,
+      "loss": 0.0307,
+      "step": 1483
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2707420067553595e-06,
+      "loss": 0.0363,
+      "step": 1484
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.257921279692372e-06,
+      "loss": 0.0421,
+      "step": 1485
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.2451077011096726e-06,
+      "loss": 0.044,
+      "step": 1486
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.232301295485004e-06,
+      "loss": 0.023,
+      "step": 1487
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.219502087282391e-06,
+      "loss": 0.0266,
+      "step": 1488
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.206710100952124e-06,
+      "loss": 0.0317,
+      "step": 1489
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.193925360930691e-06,
+      "loss": 0.0287,
+      "step": 1490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.181147891640733e-06,
+      "loss": 0.0362,
+      "step": 1491
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.168377717491005e-06,
+      "loss": 0.0279,
+      "step": 1492
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.155614862876336e-06,
+      "loss": 0.0414,
+      "step": 1493
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 6.142859352177557e-06,
+      "loss": 0.0188,
+      "step": 1494
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.13011120976148e-06,
+      "loss": 0.0339,
+      "step": 1495
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.117370459980837e-06,
+      "loss": 0.0447,
+      "step": 1496
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.104637127174237e-06,
+      "loss": 0.0314,
+      "step": 1497
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.091911235666126e-06,
+      "loss": 0.024,
+      "step": 1498
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0791928097667256e-06,
+      "loss": 0.0269,
+      "step": 1499
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.06648187377201e-06,
+      "loss": 0.0238,
+      "step": 1500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.053778451963634e-06,
+      "loss": 0.0296,
+      "step": 1501
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.041082568608901e-06,
+      "loss": 0.0305,
+      "step": 1502
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.0283942479607104e-06,
+      "loss": 0.0295,
+      "step": 1503
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.015713514257523e-06,
+      "loss": 0.0292,
+      "step": 1504
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 6.003040391723294e-06,
+      "loss": 0.0453,
+      "step": 1505
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.990374904567458e-06,
+      "loss": 0.0326,
+      "step": 1506
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.977717076984844e-06,
+      "loss": 0.0247,
+      "step": 1507
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.965066933155657e-06,
+      "loss": 0.0337,
+      "step": 1508
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.952424497245428e-06,
+      "loss": 0.0278,
+      "step": 1509
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.939789793404957e-06,
+      "loss": 0.0287,
+      "step": 1510
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.927162845770278e-06,
+      "loss": 0.0303,
+      "step": 1511
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.914543678462599e-06,
+      "loss": 0.0457,
+      "step": 1512
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.901932315588281e-06,
+      "loss": 0.0239,
+      "step": 1513
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8893287812387665e-06,
+      "loss": 0.022,
+      "step": 1514
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.876733099490542e-06,
+      "loss": 0.0268,
+      "step": 1515
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8641452944050966e-06,
+      "loss": 0.0346,
+      "step": 1516
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.851565390028872e-06,
+      "loss": 0.0387,
+      "step": 1517
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.838993410393211e-06,
+      "loss": 0.0399,
+      "step": 1518
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.826429379514333e-06,
+      "loss": 0.034,
+      "step": 1519
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.81387332139326e-06,
+      "loss": 0.0337,
+      "step": 1520
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8013252600157866e-06,
+      "loss": 0.0196,
+      "step": 1521
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.788785219352433e-06,
+      "loss": 0.0212,
+      "step": 1522
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.77625322335839e-06,
+      "loss": 0.0472,
+      "step": 1523
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.763729295973497e-06,
+      "loss": 0.0364,
+      "step": 1524
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.751213461122165e-06,
+      "loss": 0.0331,
+      "step": 1525
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.738705742713347e-06,
+      "loss": 0.0411,
+      "step": 1526
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.7262061646405045e-06,
+      "loss": 0.0253,
+      "step": 1527
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.713714750781534e-06,
+      "loss": 0.0336,
+      "step": 1528
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.70123152499874e-06,
+      "loss": 0.0252,
+      "step": 1529
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.688756511138789e-06,
+      "loss": 0.0283,
+      "step": 1530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6762897330326515e-06,
+      "loss": 0.0336,
+      "step": 1531
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6638312144955785e-06,
+      "loss": 0.0351,
+      "step": 1532
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.651380979327035e-06,
+      "loss": 0.0274,
+      "step": 1533
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.638939051310661e-06,
+      "loss": 0.025,
+      "step": 1534
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.626505454214231e-06,
+      "loss": 0.0256,
+      "step": 1535
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6140802117895995e-06,
+      "loss": 0.0375,
+      "step": 1536
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.601663347772669e-06,
+      "loss": 0.0275,
+      "step": 1537
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.589254885883326e-06,
+      "loss": 0.0335,
+      "step": 1538
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.576854849825424e-06,
+      "loss": 0.0365,
+      "step": 1539
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.564463263286704e-06,
+      "loss": 0.0405,
+      "step": 1540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.552080149938772e-06,
+      "loss": 0.0379,
+      "step": 1541
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.539705533437047e-06,
+      "loss": 0.0238,
+      "step": 1542
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.527339437420718e-06,
+      "loss": 0.0284,
+      "step": 1543
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.514981885512691e-06,
+      "loss": 0.0474,
+      "step": 1544
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.502632901319565e-06,
+      "loss": 0.0298,
+      "step": 1545
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.490292508431558e-06,
+      "loss": 0.0213,
+      "step": 1546
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.477960730422476e-06,
+      "loss": 0.0279,
+      "step": 1547
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.465637590849682e-06,
+      "loss": 0.0326,
+      "step": 1548
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.453323113254021e-06,
+      "loss": 0.0274,
+      "step": 1549
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.441017321159802e-06,
+      "loss": 0.0331,
+      "step": 1550
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.4287202380747305e-06,
+      "loss": 0.0387,
+      "step": 1551
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.416431887489892e-06,
+      "loss": 0.0311,
+      "step": 1552
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.404152292879677e-06,
+      "loss": 0.0233,
+      "step": 1553
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.391881477701753e-06,
+      "loss": 0.0298,
+      "step": 1554
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.379619465397017e-06,
+      "loss": 0.0246,
+      "step": 1555
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.367366279389549e-06,
+      "loss": 0.0321,
+      "step": 1556
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.355121943086565e-06,
+      "loss": 0.0329,
+      "step": 1557
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.342886479878388e-06,
+      "loss": 0.0261,
+      "step": 1558
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3306599131383775e-06,
+      "loss": 0.0465,
+      "step": 1559
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.3184422662229005e-06,
+      "loss": 0.0328,
+      "step": 1560
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.306233562471289e-06,
+      "loss": 0.0261,
+      "step": 1561
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.2940338252057865e-06,
+      "loss": 0.0345,
+      "step": 1562
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.281843077731512e-06,
+      "loss": 0.0278,
+      "step": 1563
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.269661343336403e-06,
+      "loss": 0.0395,
+      "step": 1564
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.257488645291193e-06,
+      "loss": 0.0304,
+      "step": 1565
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.245325006849344e-06,
+      "loss": 0.0255,
+      "step": 1566
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.233170451247012e-06,
+      "loss": 0.0306,
+      "step": 1567
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.221025001703001e-06,
+      "loss": 0.0464,
+      "step": 1568
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.208888681418727e-06,
+      "loss": 0.0363,
+      "step": 1569
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.196761513578159e-06,
+      "loss": 0.0267,
+      "step": 1570
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.184643521347781e-06,
+      "loss": 0.0217,
+      "step": 1571
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.17253472787656e-06,
+      "loss": 0.0448,
+      "step": 1572
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.16043515629588e-06,
+      "loss": 0.0194,
+      "step": 1573
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.148344829719509e-06,
+      "loss": 0.047,
+      "step": 1574
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.136263771243561e-06,
+      "loss": 0.0161,
+      "step": 1575
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.124192003946438e-06,
+      "loss": 0.0332,
+      "step": 1576
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.112129550888792e-06,
+      "loss": 0.0317,
+      "step": 1577
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.100076435113497e-06,
+      "loss": 0.0333,
+      "step": 1578
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.088032679645571e-06,
+      "loss": 0.0433,
+      "step": 1579
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.075998307492163e-06,
+      "loss": 0.0299,
+      "step": 1580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0639733416424905e-06,
+      "loss": 0.0367,
+      "step": 1581
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.051957805067806e-06,
+      "loss": 0.0265,
+      "step": 1582
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.03995172072135e-06,
+      "loss": 0.0276,
+      "step": 1583
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.0279551115383e-06,
+      "loss": 0.0274,
+      "step": 1584
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.015968000435743e-06,
+      "loss": 0.0266,
+      "step": 1585
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.003990410312616e-06,
+      "loss": 0.0282,
+      "step": 1586
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.9920223640496675e-06,
+      "loss": 0.0281,
+      "step": 1587
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.980063884509415e-06,
+      "loss": 0.0429,
+      "step": 1588
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9681149945361e-06,
+      "loss": 0.0288,
+      "step": 1589
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.956175716955651e-06,
+      "loss": 0.0327,
+      "step": 1590
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.9442460745756284e-06,
+      "loss": 0.0382,
+      "step": 1591
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.932326090185182e-06,
+      "loss": 0.0374,
+      "step": 1592
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.920415786555025e-06,
+      "loss": 0.0175,
+      "step": 1593
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.908515186437368e-06,
+      "loss": 0.025,
+      "step": 1594
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8966243125658866e-06,
+      "loss": 0.0382,
+      "step": 1595
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.884743187655675e-06,
+      "loss": 0.0317,
+      "step": 1596
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.8728718344032034e-06,
+      "loss": 0.031,
+      "step": 1597
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.861010275486286e-06,
+      "loss": 0.0319,
+      "step": 1598
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.849158533564013e-06,
+      "loss": 0.0234,
+      "step": 1599
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.837316631276728e-06,
+      "loss": 0.0431,
+      "step": 1600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.825484591245977e-06,
+      "loss": 0.0276,
+      "step": 1601
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.813662436074465e-06,
+      "loss": 0.0323,
+      "step": 1602
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.801850188346012e-06,
+      "loss": 0.025,
+      "step": 1603
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.790047870625523e-06,
+      "loss": 0.042,
+      "step": 1604
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.7782555054589185e-06,
+      "loss": 0.0208,
+      "step": 1605
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.766473115373117e-06,
+      "loss": 0.035,
+      "step": 1606
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.754700722875978e-06,
+      "loss": 0.0365,
+      "step": 1607
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.742938350456261e-06,
+      "loss": 0.0272,
+      "step": 1608
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.731186020583588e-06,
+      "loss": 0.0232,
+      "step": 1609
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.71944375570839e-06,
+      "loss": 0.0259,
+      "step": 1610
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.707711578261886e-06,
+      "loss": 0.0351,
+      "step": 1611
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.695989510656006e-06,
+      "loss": 0.0368,
+      "step": 1612
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.684277575283377e-06,
+      "loss": 0.0168,
+      "step": 1613
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.672575794517274e-06,
+      "loss": 0.0257,
+      "step": 1614
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.660884190711566e-06,
+      "loss": 0.0398,
+      "step": 1615
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.649202786200682e-06,
+      "loss": 0.0327,
+      "step": 1616
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.637531603299576e-06,
+      "loss": 0.0283,
+      "step": 1617
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.625870664303663e-06,
+      "loss": 0.0377,
+      "step": 1618
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.614219991488799e-06,
+      "loss": 0.0506,
+      "step": 1619
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.602579607111221e-06,
+      "loss": 0.0335,
+      "step": 1620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.590949533407521e-06,
+      "loss": 0.0207,
+      "step": 1621
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.579329792594585e-06,
+      "loss": 0.0241,
+      "step": 1622
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.56772040686956e-06,
+      "loss": 0.0379,
+      "step": 1623
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.556121398409828e-06,
+      "loss": 0.035,
+      "step": 1624
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.544532789372929e-06,
+      "loss": 0.0311,
+      "step": 1625
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.532954601896543e-06,
+      "loss": 0.0435,
+      "step": 1626
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.521386858098443e-06,
+      "loss": 0.0394,
+      "step": 1627
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.509829580076452e-06,
+      "loss": 0.0282,
+      "step": 1628
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.4982827899083945e-06,
+      "loss": 0.0323,
+      "step": 1629
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.486746509652071e-06,
+      "loss": 0.0297,
+      "step": 1630
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.475220761345196e-06,
+      "loss": 0.0372,
+      "step": 1631
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.463705567005367e-06,
+      "loss": 0.0288,
+      "step": 1632
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.452200948630021e-06,
+      "loss": 0.0342,
+      "step": 1633
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.440706928196387e-06,
+      "loss": 0.0361,
+      "step": 1634
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.429223527661463e-06,
+      "loss": 0.0402,
+      "step": 1635
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.4177507689619385e-06,
+      "loss": 0.0375,
+      "step": 1636
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.406288674014196e-06,
+      "loss": 0.0297,
+      "step": 1637
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.394837264714234e-06,
+      "loss": 0.0282,
+      "step": 1638
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.383396562937636e-06,
+      "loss": 0.0223,
+      "step": 1639
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.371966590539541e-06,
+      "loss": 0.0235,
+      "step": 1640
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.360547369354584e-06,
+      "loss": 0.0372,
+      "step": 1641
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.349138921196862e-06,
+      "loss": 0.0379,
+      "step": 1642
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.337741267859901e-06,
+      "loss": 0.0332,
+      "step": 1643
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.326354431116596e-06,
+      "loss": 0.036,
+      "step": 1644
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.314978432719183e-06,
+      "loss": 0.0273,
+      "step": 1645
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.303613294399194e-06,
+      "loss": 0.0394,
+      "step": 1646
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2922590378674125e-06,
+      "loss": 0.0309,
+      "step": 1647
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2809156848138366e-06,
+      "loss": 0.0294,
+      "step": 1648
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.269583256907632e-06,
+      "loss": 0.0396,
+      "step": 1649
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.258261775797106e-06,
+      "loss": 0.0208,
+      "step": 1650
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.246951263109639e-06,
+      "loss": 0.0325,
+      "step": 1651
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.235651740451667e-06,
+      "loss": 0.0274,
+      "step": 1652
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.224363229408629e-06,
+      "loss": 0.0266,
+      "step": 1653
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213085751544926e-06,
+      "loss": 0.0271,
+      "step": 1654
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.201819328403894e-06,
+      "loss": 0.032,
+      "step": 1655
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1905639815077385e-06,
+      "loss": 0.0246,
+      "step": 1656
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.1793197323575055e-06,
+      "loss": 0.0291,
+      "step": 1657
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.168086602433056e-06,
+      "loss": 0.0333,
+      "step": 1658
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.156864613192993e-06,
+      "loss": 0.0261,
+      "step": 1659
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.145653786074648e-06,
+      "loss": 0.0277,
+      "step": 1660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.134454142494023e-06,
+      "loss": 0.0294,
+      "step": 1661
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.123265703845759e-06,
+      "loss": 0.0316,
+      "step": 1662
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.112088491503096e-06,
+      "loss": 0.0208,
+      "step": 1663
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.100922526817824e-06,
+      "loss": 0.0244,
+      "step": 1664
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.089767831120245e-06,
+      "loss": 0.0272,
+      "step": 1665
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.078624425719139e-06,
+      "loss": 0.0258,
+      "step": 1666
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.067492331901712e-06,
+      "loss": 0.0222,
+      "step": 1667
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.056371570933566e-06,
+      "loss": 0.0333,
+      "step": 1668
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.0452621640586575e-06,
+      "loss": 0.0305,
+      "step": 1669
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.034164132499246e-06,
+      "loss": 0.0304,
+      "step": 1670
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.023077497455862e-06,
+      "loss": 0.0227,
+      "step": 1671
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.012002280107268e-06,
+      "loss": 0.027,
+      "step": 1672
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.000938501610414e-06,
+      "loss": 0.038,
+      "step": 1673
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.989886183100399e-06,
+      "loss": 0.0273,
+      "step": 1674
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9788453456904254e-06,
+      "loss": 0.0293,
+      "step": 1675
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.967816010471776e-06,
+      "loss": 0.0217,
+      "step": 1676
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.956798198513746e-06,
+      "loss": 0.0407,
+      "step": 1677
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.945791930863623e-06,
+      "loss": 0.0241,
+      "step": 1678
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.934797228546649e-06,
+      "loss": 0.0353,
+      "step": 1679
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.9238141125659645e-06,
+      "loss": 0.0263,
+      "step": 1680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.912842603902577e-06,
+      "loss": 0.0342,
+      "step": 1681
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.901882723515319e-06,
+      "loss": 0.0323,
+      "step": 1682
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.89093449234082e-06,
+      "loss": 0.022,
+      "step": 1683
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8799979312934464e-06,
+      "loss": 0.029,
+      "step": 1684
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.869073061265272e-06,
+      "loss": 0.0356,
+      "step": 1685
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.85815990312604e-06,
+      "loss": 0.021,
+      "step": 1686
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.847258477723119e-06,
+      "loss": 0.025,
+      "step": 1687
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.836368805881462e-06,
+      "loss": 0.0337,
+      "step": 1688
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.825490908403579e-06,
+      "loss": 0.0211,
+      "step": 1689
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.814624806069477e-06,
+      "loss": 0.0206,
+      "step": 1690
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.8037705196366336e-06,
+      "loss": 0.0257,
+      "step": 1691
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.7929280698399575e-06,
+      "loss": 0.0275,
+      "step": 1692
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.782097477391742e-06,
+      "loss": 0.0288,
+      "step": 1693
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7712787629816316e-06,
+      "loss": 0.0211,
+      "step": 1694
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.760471947276576e-06,
+      "loss": 0.0413,
+      "step": 1695
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.749677050920807e-06,
+      "loss": 0.0235,
+      "step": 1696
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.738894094535773e-06,
+      "loss": 0.0378,
+      "step": 1697
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.72812309872012e-06,
+      "loss": 0.0219,
+      "step": 1698
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7173640840496407e-06,
+      "loss": 0.0433,
+      "step": 1699
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.7066170710772532e-06,
+      "loss": 0.0269,
+      "step": 1700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.695882080332934e-06,
+      "loss": 0.027,
+      "step": 1701
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6851591323236977e-06,
+      "loss": 0.0272,
+      "step": 1702
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.674448247533562e-06,
+      "loss": 0.0258,
+      "step": 1703
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.663749446423489e-06,
+      "loss": 0.0209,
+      "step": 1704
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6530627494313643e-06,
+      "loss": 0.037,
+      "step": 1705
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6423881769719482e-06,
+      "loss": 0.0365,
+      "step": 1706
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.6317257494368407e-06,
+      "loss": 0.0327,
+      "step": 1707
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.621075487194435e-06,
+      "loss": 0.0361,
+      "step": 1708
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.610437410589902e-06,
+      "loss": 0.0357,
+      "step": 1709
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5998115399451163e-06,
+      "loss": 0.0204,
+      "step": 1710
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.589197895558645e-06,
+      "loss": 0.0204,
+      "step": 1711
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.578596497705695e-06,
+      "loss": 0.0299,
+      "step": 1712
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.568007366638082e-06,
+      "loss": 0.0292,
+      "step": 1713
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5574305225841834e-06,
+      "loss": 0.0373,
+      "step": 1714
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5468659857489152e-06,
+      "loss": 0.0323,
+      "step": 1715
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5363137763136723e-06,
+      "loss": 0.0359,
+      "step": 1716
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.525773914436307e-06,
+      "loss": 0.0358,
+      "step": 1717
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5152464202510784e-06,
+      "loss": 0.0197,
+      "step": 1718
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5047313138686256e-06,
+      "loss": 0.0285,
+      "step": 1719
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4942286153759154e-06,
+      "loss": 0.0288,
+      "step": 1720
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4837383448362248e-06,
+      "loss": 0.036,
+      "step": 1721
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4732605222890792e-06,
+      "loss": 0.0296,
+      "step": 1722
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4627951677502235e-06,
+      "loss": 0.0296,
+      "step": 1723
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.452342301211595e-06,
+      "loss": 0.0364,
+      "step": 1724
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.441901942641268e-06,
+      "loss": 0.0425,
+      "step": 1725
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4314741119834217e-06,
+      "loss": 0.0251,
+      "step": 1726
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4210588291583035e-06,
+      "loss": 0.0302,
+      "step": 1727
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.4106561140621985e-06,
+      "loss": 0.0305,
+      "step": 1728
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.400265986567376e-06,
+      "loss": 0.0276,
+      "step": 1729
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.389888466522061e-06,
+      "loss": 0.031,
+      "step": 1730
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3795235737503952e-06,
+      "loss": 0.0291,
+      "step": 1731
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3691713280523975e-06,
+      "loss": 0.037,
+      "step": 1732
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3588317492039268e-06,
+      "loss": 0.0309,
+      "step": 1733
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3485048569566446e-06,
+      "loss": 0.0356,
+      "step": 1734
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.338190671037982e-06,
+      "loss": 0.0368,
+      "step": 1735
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.327889211151093e-06,
+      "loss": 0.0254,
+      "step": 1736
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3176004969748188e-06,
+      "loss": 0.028,
+      "step": 1737
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.3073245481636575e-06,
+      "loss": 0.0194,
+      "step": 1738
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2970613843477175e-06,
+      "loss": 0.0331,
+      "step": 1739
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.2868110251326836e-06,
+      "loss": 0.0313,
+      "step": 1740
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.276573490099788e-06,
+      "loss": 0.0359,
+      "step": 1741
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2663487988057563e-06,
+      "loss": 0.0209,
+      "step": 1742
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2561369707827826e-06,
+      "loss": 0.0298,
+      "step": 1743
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.245938025538482e-06,
+      "loss": 0.0239,
+      "step": 1744
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2357519825558735e-06,
+      "loss": 0.0189,
+      "step": 1745
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.225578861293316e-06,
+      "loss": 0.025,
+      "step": 1746
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2154186811844855e-06,
+      "loss": 0.0341,
+      "step": 1747
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.2052714616383462e-06,
+      "loss": 0.0224,
+      "step": 1748
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.195137222039095e-06,
+      "loss": 0.0268,
+      "step": 1749
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1850159817461334e-06,
+      "loss": 0.0393,
+      "step": 1750
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.174907760094034e-06,
+      "loss": 0.0429,
+      "step": 1751
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.1648125763924973e-06,
+      "loss": 0.0493,
+      "step": 1752
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1547304499263164e-06,
+      "loss": 0.0257,
+      "step": 1753
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.144661399955349e-06,
+      "loss": 0.0157,
+      "step": 1754
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.134605445714465e-06,
+      "loss": 0.0233,
+      "step": 1755
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.124562606413517e-06,
+      "loss": 0.0369,
+      "step": 1756
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1145329012373092e-06,
+      "loss": 0.0338,
+      "step": 1757
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.1045163493455534e-06,
+      "loss": 0.0356,
+      "step": 1758
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0945129698728337e-06,
+      "loss": 0.0243,
+      "step": 1759
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0845227819285694e-06,
+      "loss": 0.032,
+      "step": 1760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0745458045969876e-06,
+      "loss": 0.0384,
+      "step": 1761
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.0645820569370742e-06,
+      "loss": 0.0304,
+      "step": 1762
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.054631557982539e-06,
+      "loss": 0.0273,
+      "step": 1763
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.044694326741786e-06,
+      "loss": 0.0341,
+      "step": 1764
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.03477038219788e-06,
+      "loss": 0.0367,
+      "step": 1765
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0248597433084947e-06,
+      "loss": 0.0277,
+      "step": 1766
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0149624290058864e-06,
+      "loss": 0.0436,
+      "step": 1767
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.0050784581968685e-06,
+      "loss": 0.0232,
+      "step": 1768
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9952078497627533e-06,
+      "loss": 0.0196,
+      "step": 1769
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.985350622559331e-06,
+      "loss": 0.0332,
+      "step": 1770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9755067954168295e-06,
+      "loss": 0.0246,
+      "step": 1771
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9656763871398763e-06,
+      "loss": 0.0251,
+      "step": 1772
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9558594165074673e-06,
+      "loss": 0.0258,
+      "step": 1773
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9460559022729318e-06,
+      "loss": 0.035,
+      "step": 1774
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.9362658631638874e-06,
+      "loss": 0.0355,
+      "step": 1775
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.9264893178822127e-06,
+      "loss": 0.0353,
+      "step": 1776
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.91672628510401e-06,
+      "loss": 0.0263,
+      "step": 1777
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.906976783479566e-06,
+      "loss": 0.0251,
+      "step": 1778
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.897240831633321e-06,
+      "loss": 0.0449,
+      "step": 1779
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8875184481638283e-06,
+      "loss": 0.0312,
+      "step": 1780
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8778096516437304e-06,
+      "loss": 0.0389,
+      "step": 1781
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8681144606197055e-06,
+      "loss": 0.0298,
+      "step": 1782
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8584328936124427e-06,
+      "loss": 0.0296,
+      "step": 1783
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8487649691166077e-06,
+      "loss": 0.0234,
+      "step": 1784
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8391107056007994e-06,
+      "loss": 0.0196,
+      "step": 1785
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.8294701215075338e-06,
+      "loss": 0.0165,
+      "step": 1786
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.81984323525318e-06,
+      "loss": 0.0301,
+      "step": 1787
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8102300652279445e-06,
+      "loss": 0.0329,
+      "step": 1788
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.8006306297958395e-06,
+      "loss": 0.0252,
+      "step": 1789
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.79104494729463e-06,
+      "loss": 0.0255,
+      "step": 1790
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7814730360358142e-06,
+      "loss": 0.027,
+      "step": 1791
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.771914914304581e-06,
+      "loss": 0.0106,
+      "step": 1792
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.762370600359774e-06,
+      "loss": 0.037,
+      "step": 1793
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.752840112433871e-06,
+      "loss": 0.0194,
+      "step": 1794
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.743323468732928e-06,
+      "loss": 0.029,
+      "step": 1795
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.733820687436556e-06,
+      "loss": 0.0225,
+      "step": 1796
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.724331786697886e-06,
+      "loss": 0.0256,
+      "step": 1797
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7148567846435337e-06,
+      "loss": 0.0414,
+      "step": 1798
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.7053956993735607e-06,
+      "loss": 0.02,
+      "step": 1799
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6959485489614512e-06,
+      "loss": 0.0324,
+      "step": 1800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.686515351454062e-06,
+      "loss": 0.0246,
+      "step": 1801
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.677096124871598e-06,
+      "loss": 0.0221,
+      "step": 1802
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6676908872075756e-06,
+      "loss": 0.0296,
+      "step": 1803
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.65829965642879e-06,
+      "loss": 0.0264,
+      "step": 1804
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6489224504752753e-06,
+      "loss": 0.0249,
+      "step": 1805
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.639559287260275e-06,
+      "loss": 0.0327,
+      "step": 1806
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6302101846702135e-06,
+      "loss": 0.0208,
+      "step": 1807
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6208751605646456e-06,
+      "loss": 0.0366,
+      "step": 1808
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6115542327762342e-06,
+      "loss": 0.0267,
+      "step": 1809
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.602247419110723e-06,
+      "loss": 0.0209,
+      "step": 1810
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.592954737346882e-06,
+      "loss": 0.039,
+      "step": 1811
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5836762052364873e-06,
+      "loss": 0.0367,
+      "step": 1812
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.574411840504293e-06,
+      "loss": 0.0376,
+      "step": 1813
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.565161660847981e-06,
+      "loss": 0.0328,
+      "step": 1814
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.55592568393814e-06,
+      "loss": 0.0292,
+      "step": 1815
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5467039274182213e-06,
+      "loss": 0.0302,
+      "step": 1816
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.537496408904517e-06,
+      "loss": 0.0261,
+      "step": 1817
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5283031459861207e-06,
+      "loss": 0.0299,
+      "step": 1818
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.519124156224885e-06,
+      "loss": 0.0352,
+      "step": 1819
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.509959457155411e-06,
+      "loss": 0.0373,
+      "step": 1820
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5008090662849884e-06,
+      "loss": 0.0214,
+      "step": 1821
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4916730010935785e-06,
+      "loss": 0.0256,
+      "step": 1822
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4825512790337747e-06,
+      "loss": 0.0366,
+      "step": 1823
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4734439175307723e-06,
+      "loss": 0.036,
+      "step": 1824
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4643509339823302e-06,
+      "loss": 0.0373,
+      "step": 1825
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4552723457587492e-06,
+      "loss": 0.0281,
+      "step": 1826
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.446208170202824e-06,
+      "loss": 0.0231,
+      "step": 1827
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4371584246298174e-06,
+      "loss": 0.0314,
+      "step": 1828
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4281231263274286e-06,
+      "loss": 0.0375,
+      "step": 1829
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.419102292555754e-06,
+      "loss": 0.0245,
+      "step": 1830
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4100959405472656e-06,
+      "loss": 0.0157,
+      "step": 1831
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4011040875067682e-06,
+      "loss": 0.0282,
+      "step": 1832
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3921267506113623e-06,
+      "loss": 0.0292,
+      "step": 1833
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3831639470104296e-06,
+      "loss": 0.0376,
+      "step": 1834
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3742156938255824e-06,
+      "loss": 0.0209,
+      "step": 1835
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3652820081506376e-06,
+      "loss": 0.0292,
+      "step": 1836
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.356362907051585e-06,
+      "loss": 0.0214,
+      "step": 1837
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.34745840756655e-06,
+      "loss": 0.0217,
+      "step": 1838
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3385685267057723e-06,
+      "loss": 0.0413,
+      "step": 1839
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3296932814515605e-06,
+      "loss": 0.028,
+      "step": 1840
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3208326887582646e-06,
+      "loss": 0.0467,
+      "step": 1841
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3119867655522445e-06,
+      "loss": 0.0276,
+      "step": 1842
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3031555287318376e-06,
+      "loss": 0.0321,
+      "step": 1843
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2943389951673257e-06,
+      "loss": 0.041,
+      "step": 1844
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.2855371817009e-06,
+      "loss": 0.0207,
+      "step": 1845
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.27675010514664e-06,
+      "loss": 0.0268,
+      "step": 1846
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.267977782290467e-06,
+      "loss": 0.0266,
+      "step": 1847
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2592202298901176e-06,
+      "loss": 0.0334,
+      "step": 1848
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.250477464675116e-06,
+      "loss": 0.0199,
+      "step": 1849
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.241749503346736e-06,
+      "loss": 0.0285,
+      "step": 1850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2330363625779704e-06,
+      "loss": 0.0397,
+      "step": 1851
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2243380590135078e-06,
+      "loss": 0.0286,
+      "step": 1852
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.215654609269685e-06,
+      "loss": 0.0335,
+      "step": 1853
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.2069860299344635e-06,
+      "loss": 0.0249,
+      "step": 1854
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.198332337567408e-06,
+      "loss": 0.0359,
+      "step": 1855
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.189693548699634e-06,
+      "loss": 0.0363,
+      "step": 1856
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1810696798337904e-06,
+      "loss": 0.0348,
+      "step": 1857
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.172460747444022e-06,
+      "loss": 0.0323,
+      "step": 1858
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1638667679759484e-06,
+      "loss": 0.0126,
+      "step": 1859
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.155287757846617e-06,
+      "loss": 0.0316,
+      "step": 1860
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1467237334444824e-06,
+      "loss": 0.022,
+      "step": 1861
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1381747111293703e-06,
+      "loss": 0.019,
+      "step": 1862
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.12964070723245e-06,
+      "loss": 0.0435,
+      "step": 1863
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1211217380561966e-06,
+      "loss": 0.0266,
+      "step": 1864
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.112617819874374e-06,
+      "loss": 0.0319,
+      "step": 1865
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1041289689319856e-06,
+      "loss": 0.0374,
+      "step": 1866
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0956552014452536e-06,
+      "loss": 0.0446,
+      "step": 1867
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0871965336015888e-06,
+      "loss": 0.0366,
+      "step": 1868
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.0787529815595554e-06,
+      "loss": 0.0302,
+      "step": 1869
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.070324561448841e-06,
+      "loss": 0.0226,
+      "step": 1870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0619112893702253e-06,
+      "loss": 0.0238,
+      "step": 1871
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.053513181395561e-06,
+      "loss": 0.0366,
+      "step": 1872
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.045130253567721e-06,
+      "loss": 0.0304,
+      "step": 1873
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0367625219005803e-06,
+      "loss": 0.0183,
+      "step": 1874
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0284100023789964e-06,
+      "loss": 0.0307,
+      "step": 1875
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.020072710958754e-06,
+      "loss": 0.031,
+      "step": 1876
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.0117506635665505e-06,
+      "loss": 0.0242,
+      "step": 1877
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.00344387609997e-06,
+      "loss": 0.0236,
+      "step": 1878
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9951523644274386e-06,
+      "loss": 0.0311,
+      "step": 1879
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9868761443882022e-06,
+      "loss": 0.0234,
+      "step": 1880
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9786152317922967e-06,
+      "loss": 0.0306,
+      "step": 1881
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.970369642420514e-06,
+      "loss": 0.0391,
+      "step": 1882
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.962139392024377e-06,
+      "loss": 0.0265,
+      "step": 1883
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9539244963261017e-06,
+      "loss": 0.0266,
+      "step": 1884
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.945724971018581e-06,
+      "loss": 0.0346,
+      "step": 1885
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9375408317653375e-06,
+      "loss": 0.0384,
+      "step": 1886
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9293720942005044e-06,
+      "loss": 0.0221,
+      "step": 1887
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9212187739287946e-06,
+      "loss": 0.0195,
+      "step": 1888
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.913080886525467e-06,
+      "loss": 0.0337,
+      "step": 1889
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9049584475362992e-06,
+      "loss": 0.0445,
+      "step": 1890
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8968514724775588e-06,
+      "loss": 0.0182,
+      "step": 1891
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8887599768359776e-06,
+      "loss": 0.0284,
+      "step": 1892
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8806839760687079e-06,
+      "loss": 0.0216,
+      "step": 1893
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8726234856033077e-06,
+      "loss": 0.0273,
+      "step": 1894
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8645785208377022e-06,
+      "loss": 0.0193,
+      "step": 1895
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8565490971401645e-06,
+      "loss": 0.0291,
+      "step": 1896
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.848535229849273e-06,
+      "loss": 0.0278,
+      "step": 1897
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.840536934273891e-06,
+      "loss": 0.0283,
+      "step": 1898
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8325542256931373e-06,
+      "loss": 0.0256,
+      "step": 1899
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8245871193563525e-06,
+      "loss": 0.0276,
+      "step": 1900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8166356304830724e-06,
+      "loss": 0.032,
+      "step": 1901
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8086997742629997e-06,
+      "loss": 0.0232,
+      "step": 1902
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8007795658559713e-06,
+      "loss": 0.0367,
+      "step": 1903
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.7928750203919353e-06,
+      "loss": 0.0195,
+      "step": 1904
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.78498615297092e-06,
+      "loss": 0.0267,
+      "step": 1905
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7771129786630025e-06,
+      "loss": 0.0184,
+      "step": 1906
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7692555125082788e-06,
+      "loss": 0.0241,
+      "step": 1907
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.761413769516841e-06,
+      "loss": 0.0262,
+      "step": 1908
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7535877646687428e-06,
+      "loss": 0.0182,
+      "step": 1909
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7457775129139742e-06,
+      "loss": 0.0337,
+      "step": 1910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7379830291724365e-06,
+      "loss": 0.0314,
+      "step": 1911
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.730204328333902e-06,
+      "loss": 0.0445,
+      "step": 1912
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.722441425257999e-06,
+      "loss": 0.02,
+      "step": 1913
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.714694334774173e-06,
+      "loss": 0.0224,
+      "step": 1914
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.7069630716816653e-06,
+      "loss": 0.0308,
+      "step": 1915
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.6992476507494793e-06,
+      "loss": 0.0329,
+      "step": 1916
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6915480867163624e-06,
+      "loss": 0.0272,
+      "step": 1917
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6838643942907628e-06,
+      "loss": 0.0151,
+      "step": 1918
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6761965881508092e-06,
+      "loss": 0.0197,
+      "step": 1919
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6685446829442919e-06,
+      "loss": 0.028,
+      "step": 1920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6609086932886165e-06,
+      "loss": 0.0282,
+      "step": 1921
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6532886337707876e-06,
+      "loss": 0.032,
+      "step": 1922
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6456845189473768e-06,
+      "loss": 0.0242,
+      "step": 1923
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6380963633445034e-06,
+      "loss": 0.0259,
+      "step": 1924
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6305241814577942e-06,
+      "loss": 0.028,
+      "step": 1925
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.622967987752362e-06,
+      "loss": 0.0247,
+      "step": 1926
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.6154277966627773e-06,
+      "loss": 0.0242,
+      "step": 1927
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.6079036225930422e-06,
+      "loss": 0.0276,
+      "step": 1928
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.600395479916561e-06,
+      "loss": 0.0245,
+      "step": 1929
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5929033829761108e-06,
+      "loss": 0.0179,
+      "step": 1930
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5854273460838232e-06,
+      "loss": 0.0309,
+      "step": 1931
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.577967383521144e-06,
+      "loss": 0.0262,
+      "step": 1932
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5705235095388138e-06,
+      "loss": 0.0344,
+      "step": 1933
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5630957383568413e-06,
+      "loss": 0.0303,
+      "step": 1934
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5556840841644716e-06,
+      "loss": 0.0252,
+      "step": 1935
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5482885611201603e-06,
+      "loss": 0.0287,
+      "step": 1936
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5409091833515554e-06,
+      "loss": 0.023,
+      "step": 1937
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.533545964955454e-06,
+      "loss": 0.0264,
+      "step": 1938
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5261989199977867e-06,
+      "loss": 0.0226,
+      "step": 1939
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5188680625135889e-06,
+      "loss": 0.0188,
+      "step": 1940
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5115534065069736e-06,
+      "loss": 0.0244,
+      "step": 1941
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.504254965951104e-06,
+      "loss": 0.0303,
+      "step": 1942
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.496972754788163e-06,
+      "loss": 0.0279,
+      "step": 1943
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.489706786929339e-06,
+      "loss": 0.0236,
+      "step": 1944
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.482457076254783e-06,
+      "loss": 0.0298,
+      "step": 1945
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.475223636613592e-06,
+      "loss": 0.0258,
+      "step": 1946
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4680064818237826e-06,
+      "loss": 0.0156,
+      "step": 1947
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4608056256722602e-06,
+      "loss": 0.0166,
+      "step": 1948
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4536210819147944e-06,
+      "loss": 0.0226,
+      "step": 1949
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4464528642759989e-06,
+      "loss": 0.0195,
+      "step": 1950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4393009864492924e-06,
+      "loss": 0.0309,
+      "step": 1951
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.432165462096885e-06,
+      "loss": 0.0278,
+      "step": 1952
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4250463048497421e-06,
+      "loss": 0.0385,
+      "step": 1953
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4179435283075682e-06,
+      "loss": 0.0325,
+      "step": 1954
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.410857146038771e-06,
+      "loss": 0.0271,
+      "step": 1955
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4037871715804407e-06,
+      "loss": 0.0211,
+      "step": 1956
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3967336184383318e-06,
+      "loss": 0.0462,
+      "step": 1957
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.389696500086819e-06,
+      "loss": 0.0221,
+      "step": 1958
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3826758299688858e-06,
+      "loss": 0.0241,
+      "step": 1959
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3756716214960952e-06,
+      "loss": 0.0199,
+      "step": 1960
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.368683888048561e-06,
+      "loss": 0.0217,
+      "step": 1961
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3617126429749293e-06,
+      "loss": 0.0304,
+      "step": 1962
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.354757899592345e-06,
+      "loss": 0.0386,
+      "step": 1963
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.347819671186429e-06,
+      "loss": 0.0419,
+      "step": 1964
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.340897971011259e-06,
+      "loss": 0.0299,
+      "step": 1965
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3339928122893331e-06,
+      "loss": 0.0262,
+      "step": 1966
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3271042082115537e-06,
+      "loss": 0.0396,
+      "step": 1967
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3202321719371968e-06,
+      "loss": 0.0303,
+      "step": 1968
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3133767165938883e-06,
+      "loss": 0.0239,
+      "step": 1969
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3065378552775853e-06,
+      "loss": 0.0325,
+      "step": 1970
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2997156010525402e-06,
+      "loss": 0.0206,
+      "step": 1971
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2929099669512801e-06,
+      "loss": 0.0308,
+      "step": 1972
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2861209659745866e-06,
+      "loss": 0.037,
+      "step": 1973
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2793486110914643e-06,
+      "loss": 0.0289,
+      "step": 1974
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2725929152391182e-06,
+      "loss": 0.0165,
+      "step": 1975
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2658538913229357e-06,
+      "loss": 0.0293,
+      "step": 1976
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2591315522164483e-06,
+      "loss": 0.0337,
+      "step": 1977
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.252425910761318e-06,
+      "loss": 0.0416,
+      "step": 1978
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.245736979767308e-06,
+      "loss": 0.0345,
+      "step": 1979
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2390647720122627e-06,
+      "loss": 0.0302,
+      "step": 1980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2324093002420756e-06,
+      "loss": 0.043,
+      "step": 1981
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.225770577170672e-06,
+      "loss": 0.0257,
+      "step": 1982
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2191486154799847e-06,
+      "loss": 0.027,
+      "step": 1983
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.21254342781992e-06,
+      "loss": 0.034,
+      "step": 1984
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.205955026808352e-06,
+      "loss": 0.0385,
+      "step": 1985
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.1993834250310766e-06,
+      "loss": 0.0302,
+      "step": 1986
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1928286350418028e-06,
+      "loss": 0.0212,
+      "step": 1987
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1862906693621234e-06,
+      "loss": 0.0226,
+      "step": 1988
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1797695404814914e-06,
+      "loss": 0.0185,
+      "step": 1989
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1732652608572003e-06,
+      "loss": 0.0263,
+      "step": 1990
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1667778429143516e-06,
+      "loss": 0.035,
+      "step": 1991
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1603072990458374e-06,
+      "loss": 0.034,
+      "step": 1992
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1538536416123169e-06,
+      "loss": 0.0202,
+      "step": 1993
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1474168829421883e-06,
+      "loss": 0.0207,
+      "step": 1994
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.140997035331569e-06,
+      "loss": 0.0256,
+      "step": 1995
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.134594111044276e-06,
+      "loss": 0.0258,
+      "step": 1996
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.1282081223117914e-06,
+      "loss": 0.0288,
+      "step": 1997
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.121839081333248e-06,
+      "loss": 0.0269,
+      "step": 1998
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1154870002754037e-06,
+      "loss": 0.0233,
+      "step": 1999
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1091518912726163e-06,
+      "loss": 0.0269,
+      "step": 2000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.1028337664268251e-06,
+      "loss": 0.0337,
+      "step": 2001
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0965326378075181e-06,
+      "loss": 0.0136,
+      "step": 2002
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0902485174517253e-06,
+      "loss": 0.028,
+      "step": 2003
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0839814173639785e-06,
+      "loss": 0.045,
+      "step": 2004
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0777313495162945e-06,
+      "loss": 0.0215,
+      "step": 2005
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0714983258481627e-06,
+      "loss": 0.0332,
+      "step": 2006
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.065282358266504e-06,
+      "loss": 0.0321,
+      "step": 2007
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.0590834586456577e-06,
+      "loss": 0.032,
+      "step": 2008
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.052901638827366e-06,
+      "loss": 0.0285,
+      "step": 2009
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0467369106207349e-06,
+      "loss": 0.0259,
+      "step": 2010
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.040589285802225e-06,
+      "loss": 0.0277,
+      "step": 2011
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0344587761156234e-06,
+      "loss": 0.0366,
+      "step": 2012
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.02834539327202e-06,
+      "loss": 0.0292,
+      "step": 2013
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.022249148949791e-06,
+      "loss": 0.029,
+      "step": 2014
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0161700547945697e-06,
+      "loss": 0.0307,
+      "step": 2015
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.010108122419231e-06,
+      "loss": 0.0346,
+      "step": 2016
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.0040633634038632e-06,
+      "loss": 0.035,
+      "step": 2017
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980357892957493e-07,
+      "loss": 0.0206,
+      "step": 2018
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.92025411609342e-07,
+      "loss": 0.0307,
+      "step": 2019
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.86032241826248e-07,
+      "loss": 0.0234,
+      "step": 2020
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.80056291395194e-07,
+      "loss": 0.0312,
+      "step": 2021
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.740975717320244e-07,
+      "loss": 0.0226,
+      "step": 2022
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.681560942196588e-07,
+      "loss": 0.0348,
+      "step": 2023
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.622318702080803e-07,
+      "loss": 0.0348,
+      "step": 2024
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.563249110143148e-07,
+      "loss": 0.0281,
+      "step": 2025
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.504352279224049e-07,
+      "loss": 0.0266,
+      "step": 2026
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.445628321833955e-07,
+      "loss": 0.0315,
+      "step": 2027
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.387077350153018e-07,
+      "loss": 0.0304,
+      "step": 2028
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.328699476030955e-07,
+      "loss": 0.0309,
+      "step": 2029
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.270494810986841e-07,
+      "loss": 0.0256,
+      "step": 2030
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.212463466208854e-07,
+      "loss": 0.0291,
+      "step": 2031
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.154605552554063e-07,
+      "loss": 0.0215,
+      "step": 2032
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.096921180548235e-07,
+      "loss": 0.0279,
+      "step": 2033
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.039410460385634e-07,
+      "loss": 0.0288,
+      "step": 2034
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.982073501928801e-07,
+      "loss": 0.0227,
+      "step": 2035
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.924910414708343e-07,
+      "loss": 0.0245,
+      "step": 2036
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.8679213079227e-07,
+      "loss": 0.0385,
+      "step": 2037
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.811106290437977e-07,
+      "loss": 0.0282,
+      "step": 2038
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.754465470787688e-07,
+      "loss": 0.0488,
+      "step": 2039
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.697998957172627e-07,
+      "loss": 0.0354,
+      "step": 2040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.641706857460541e-07,
+      "loss": 0.0311,
+      "step": 2041
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.58558927918609e-07,
+      "loss": 0.045,
+      "step": 2042
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.529646329550468e-07,
+      "loss": 0.0361,
+      "step": 2043
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.473878115421296e-07,
+      "loss": 0.028,
+      "step": 2044
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.418284743332417e-07,
+      "loss": 0.0287,
+      "step": 2045
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.362866319483653e-07,
+      "loss": 0.0265,
+      "step": 2046
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.307622949740623e-07,
+      "loss": 0.022,
+      "step": 2047
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.252554739634577e-07,
+      "loss": 0.0301,
+      "step": 2048
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.1976617943621e-07,
+      "loss": 0.0209,
+      "step": 2049
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.142944218785009e-07,
+      "loss": 0.0264,
+      "step": 2050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.088402117430116e-07,
+      "loss": 0.0236,
+      "step": 2051
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.034035594489009e-07,
+      "loss": 0.0263,
+      "step": 2052
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.979844753817856e-07,
+      "loss": 0.0232,
+      "step": 2053
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.925829698937237e-07,
+      "loss": 0.0247,
+      "step": 2054
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.871990533031948e-07,
+      "loss": 0.022,
+      "step": 2055
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.818327358950762e-07,
+      "loss": 0.0194,
+      "step": 2056
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.764840279206254e-07,
+      "loss": 0.0259,
+      "step": 2057
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.711529395974593e-07,
+      "loss": 0.0209,
+      "step": 2058
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.658394811095404e-07,
+      "loss": 0.0196,
+      "step": 2059
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.605436626071471e-07,
+      "loss": 0.0264,
+      "step": 2060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.552654942068682e-07,
+      "loss": 0.0375,
+      "step": 2061
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.500049859915681e-07,
+      "loss": 0.0246,
+      "step": 2062
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.447621480103783e-07,
+      "loss": 0.0371,
+      "step": 2063
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.395369902786742e-07,
+      "loss": 0.0327,
+      "step": 2064
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.343295227780567e-07,
+      "loss": 0.0286,
+      "step": 2065
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.29139755456334e-07,
+      "loss": 0.0229,
+      "step": 2066
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.23967698227499e-07,
+      "loss": 0.0185,
+      "step": 2067
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.188133609717185e-07,
+      "loss": 0.0365,
+      "step": 2068
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.136767535353029e-07,
+      "loss": 0.0378,
+      "step": 2069
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.085578857306974e-07,
+      "loss": 0.0239,
+      "step": 2070
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.034567673364557e-07,
+      "loss": 0.0312,
+      "step": 2071
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.983734080972304e-07,
+      "loss": 0.0192,
+      "step": 2072
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.933078177237429e-07,
+      "loss": 0.0281,
+      "step": 2073
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.882600058927724e-07,
+      "loss": 0.023,
+      "step": 2074
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.8322998224714e-07,
+      "loss": 0.0385,
+      "step": 2075
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.782177563956805e-07,
+      "loss": 0.0352,
+      "step": 2076
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.732233379132325e-07,
+      "loss": 0.0314,
+      "step": 2077
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.682467363406175e-07,
+      "loss": 0.0201,
+      "step": 2078
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.632879611846188e-07,
+      "loss": 0.031,
+      "step": 2079
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 6.583470219179677e-07,
+      "loss": 0.0235,
+      "step": 2080
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.534239279793264e-07,
+      "loss": 0.0318,
+      "step": 2081
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.485186887732631e-07,
+      "loss": 0.028,
+      "step": 2082
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.436313136702388e-07,
+      "loss": 0.0259,
+      "step": 2083
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.387618120065909e-07,
+      "loss": 0.0266,
+      "step": 2084
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.339101930845104e-07,
+      "loss": 0.031,
+      "step": 2085
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.290764661720294e-07,
+      "loss": 0.0353,
+      "step": 2086
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.242606405029993e-07,
+      "loss": 0.0317,
+      "step": 2087
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.194627252770768e-07,
+      "loss": 0.0232,
+      "step": 2088
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.146827296597035e-07,
+      "loss": 0.0287,
+      "step": 2089
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.099206627820875e-07,
+      "loss": 0.0311,
+      "step": 2090
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.051765337411886e-07,
+      "loss": 0.0199,
+      "step": 2091
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.004503515997007e-07,
+      "loss": 0.0327,
+      "step": 2092
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.957421253860352e-07,
+      "loss": 0.0231,
+      "step": 2093
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.910518640942975e-07,
+      "loss": 0.0162,
+      "step": 2094
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.863795766842817e-07,
+      "loss": 0.0245,
+      "step": 2095
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.817252720814404e-07,
+      "loss": 0.0269,
+      "step": 2096
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.770889591768757e-07,
+      "loss": 0.0291,
+      "step": 2097
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.724706468273212e-07,
+      "loss": 0.0232,
+      "step": 2098
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.678703438551236e-07,
+      "loss": 0.0195,
+      "step": 2099
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.632880590482237e-07,
+      "loss": 0.0366,
+      "step": 2100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.587238011601481e-07,
+      "loss": 0.0269,
+      "step": 2101
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.541775789099813e-07,
+      "loss": 0.0335,
+      "step": 2102
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.49649400982356e-07,
+      "loss": 0.0218,
+      "step": 2103
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.451392760274354e-07,
+      "loss": 0.0259,
+      "step": 2104
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.406472126608975e-07,
+      "loss": 0.0327,
+      "step": 2105
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.361732194639114e-07,
+      "loss": 0.0229,
+      "step": 2106
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.317173049831359e-07,
+      "loss": 0.0338,
+      "step": 2107
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.272794777306878e-07,
+      "loss": 0.017,
+      "step": 2108
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.228597461841334e-07,
+      "loss": 0.0161,
+      "step": 2109
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.18458118786469e-07,
+      "loss": 0.0301,
+      "step": 2110
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.140746039461119e-07,
+      "loss": 0.0399,
+      "step": 2111
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.097092100368727e-07,
+      "loss": 0.0213,
+      "step": 2112
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.053619453979486e-07,
+      "loss": 0.0355,
+      "step": 2113
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.010328183339075e-07,
+      "loss": 0.0305,
+      "step": 2114
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.967218371146621e-07,
+      "loss": 0.028,
+      "step": 2115
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.924290099754692e-07,
+      "loss": 0.0205,
+      "step": 2116
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.881543451169002e-07,
+      "loss": 0.0276,
+      "step": 2117
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.83897850704832e-07,
+      "loss": 0.0223,
+      "step": 2118
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.796595348704314e-07,
+      "loss": 0.0331,
+      "step": 2119
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.754394057101408e-07,
+      "loss": 0.0287,
+      "step": 2120
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.712374712856582e-07,
+      "loss": 0.0226,
+      "step": 2121
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.670537396239228e-07,
+      "loss": 0.0302,
+      "step": 2122
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.6288821871710464e-07,
+      "loss": 0.0294,
+      "step": 2123
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5874091652258315e-07,
+      "loss": 0.0307,
+      "step": 2124
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.546118409629374e-07,
+      "loss": 0.0299,
+      "step": 2125
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.5050099992592246e-07,
+      "loss": 0.0351,
+      "step": 2126
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.4640840126446873e-07,
+      "loss": 0.0298,
+      "step": 2127
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.423340527966513e-07,
+      "loss": 0.0236,
+      "step": 2128
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.382779623056843e-07,
+      "loss": 0.0231,
+      "step": 2129
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.342401375399041e-07,
+      "loss": 0.0217,
+      "step": 2130
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.30220586212754e-07,
+      "loss": 0.0244,
+      "step": 2131
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.262193160027661e-07,
+      "loss": 0.0252,
+      "step": 2132
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.222363345535585e-07,
+      "loss": 0.0294,
+      "step": 2133
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.182716494738037e-07,
+      "loss": 0.0125,
+      "step": 2134
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.1432526833722676e-07,
+      "loss": 0.0373,
+      "step": 2135
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.103971986825861e-07,
+      "loss": 0.0255,
+      "step": 2136
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0648744801366046e-07,
+      "loss": 0.0408,
+      "step": 2137
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.0259602379923323e-07,
+      "loss": 0.0251,
+      "step": 2138
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.98722933473078e-07,
+      "loss": 0.019,
+      "step": 2139
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.948681844339475e-07,
+      "loss": 0.0253,
+      "step": 2140
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9103178404555575e-07,
+      "loss": 0.0273,
+      "step": 2141
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8721373963656384e-07,
+      "loss": 0.0339,
+      "step": 2142
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8341405850056965e-07,
+      "loss": 0.0305,
+      "step": 2143
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.796327478960926e-07,
+      "loss": 0.0157,
+      "step": 2144
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.758698150465545e-07,
+      "loss": 0.0192,
+      "step": 2145
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.721252671402753e-07,
+      "loss": 0.0294,
+      "step": 2146
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.68399111330453e-07,
+      "loss": 0.0325,
+      "step": 2147
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.646913547351494e-07,
+      "loss": 0.0232,
+      "step": 2148
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.610020044372786e-07,
+      "loss": 0.0241,
+      "step": 2149
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.573310674845953e-07,
+      "loss": 0.0234,
+      "step": 2150
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.536785508896767e-07,
+      "loss": 0.0284,
+      "step": 2151
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.500444616299126e-07,
+      "loss": 0.0337,
+      "step": 2152
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.46428806647493e-07,
+      "loss": 0.0261,
+      "step": 2153
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.4283159284939083e-07,
+      "loss": 0.0428,
+      "step": 2154
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.392528271073503e-07,
+      "loss": 0.0417,
+      "step": 2155
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3569251625787725e-07,
+      "loss": 0.0334,
+      "step": 2156
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.32150667102219e-07,
+      "loss": 0.0298,
+      "step": 2157
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.286272864063611e-07,
+      "loss": 0.0226,
+      "step": 2158
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.251223809010029e-07,
+      "loss": 0.0273,
+      "step": 2159
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.216359572815542e-07,
+      "loss": 0.0252,
+      "step": 2160
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1816802220812073e-07,
+      "loss": 0.0195,
+      "step": 2161
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1471858230548433e-07,
+      "loss": 0.0261,
+      "step": 2162
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1128764416309853e-07,
+      "loss": 0.0238,
+      "step": 2163
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.078752143350738e-07,
+      "loss": 0.0234,
+      "step": 2164
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.044812993401625e-07,
+      "loss": 0.0234,
+      "step": 2165
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0110590566174936e-07,
+      "loss": 0.0279,
+      "step": 2166
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.9774903974783775e-07,
+      "loss": 0.0277,
+      "step": 2167
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.944107080110381e-07,
+      "loss": 0.0279,
+      "step": 2168
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.910909168285514e-07,
+      "loss": 0.0347,
+      "step": 2169
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.877896725421658e-07,
+      "loss": 0.021,
+      "step": 2170
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.845069814582357e-07,
+      "loss": 0.0318,
+      "step": 2171
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.81242849847676e-07,
+      "loss": 0.0376,
+      "step": 2172
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.779972839459455e-07,
+      "loss": 0.0228,
+      "step": 2173
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.74770289953038e-07,
+      "loss": 0.0271,
+      "step": 2174
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7156187403347024e-07,
+      "loss": 0.0266,
+      "step": 2175
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.683720423162661e-07,
+      "loss": 0.0199,
+      "step": 2176
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.6520080089495136e-07,
+      "loss": 0.0284,
+      "step": 2177
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.620481558275367e-07,
+      "loss": 0.022,
+      "step": 2178
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.589141131365114e-07,
+      "loss": 0.0296,
+      "step": 2179
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5579867880882517e-07,
+      "loss": 0.0163,
+      "step": 2180
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.5270185879588184e-07,
+      "loss": 0.0273,
+      "step": 2181
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4962365901352703e-07,
+      "loss": 0.0222,
+      "step": 2182
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.465640853420337e-07,
+      "loss": 0.0211,
+      "step": 2183
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4352314362609654e-07,
+      "loss": 0.0268,
+      "step": 2184
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.405008396748154e-07,
+      "loss": 0.031,
+      "step": 2185
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3749717926168759e-07,
+      "loss": 0.0275,
+      "step": 2186
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3451216812459654e-07,
+      "loss": 0.0145,
+      "step": 2187
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.315458119657965e-07,
+      "loss": 0.0185,
+      "step": 2188
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2859811645190911e-07,
+      "loss": 0.0228,
+      "step": 2189
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.256690872139067e-07,
+      "loss": 0.0316,
+      "step": 2190
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.2275872984710345e-07,
+      "loss": 0.0272,
+      "step": 2191
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.198670499111466e-07,
+      "loss": 0.0252,
+      "step": 2192
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1699405293000186e-07,
+      "loss": 0.0223,
+      "step": 2193
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1413974439194463e-07,
+      "loss": 0.0269,
+      "step": 2194
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.1130412974955218e-07,
+      "loss": 0.0354,
+      "step": 2195
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0848721441968922e-07,
+      "loss": 0.0234,
+      "step": 2196
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.056890037834991e-07,
+      "loss": 0.0214,
+      "step": 2197
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0290950318639257e-07,
+      "loss": 0.0406,
+      "step": 2198
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.0014871793804347e-07,
+      "loss": 0.0276,
+      "step": 2199
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9740665331236863e-07,
+      "loss": 0.0335,
+      "step": 2200
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9468331454752575e-07,
+      "loss": 0.0248,
+      "step": 2201
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.919787068458978e-07,
+      "loss": 0.0378,
+      "step": 2202
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.892928353740897e-07,
+      "loss": 0.019,
+      "step": 2203
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8662570526291168e-07,
+      "loss": 0.0175,
+      "step": 2204
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.839773216073737e-07,
+      "loss": 0.017,
+      "step": 2205
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8134768946667436e-07,
+      "loss": 0.0309,
+      "step": 2206
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7873681386418984e-07,
+      "loss": 0.0251,
+      "step": 2207
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7614469978746828e-07,
+      "loss": 0.034,
+      "step": 2208
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7357135218821543e-07,
+      "loss": 0.03,
+      "step": 2209
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7101677598228895e-07,
+      "loss": 0.0287,
+      "step": 2210
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6848097604968637e-07,
+      "loss": 0.0367,
+      "step": 2211
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.659639572345395e-07,
+      "loss": 0.0437,
+      "step": 2212
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6346572434509877e-07,
+      "loss": 0.022,
+      "step": 2213
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6098628215373225e-07,
+      "loss": 0.0262,
+      "step": 2214
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.585256353969067e-07,
+      "loss": 0.0296,
+      "step": 2215
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5608378877519093e-07,
+      "loss": 0.0245,
+      "step": 2216
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5366074695323253e-07,
+      "loss": 0.0239,
+      "step": 2217
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.512565145597633e-07,
+      "loss": 0.0232,
+      "step": 2218
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4887109618757946e-07,
+      "loss": 0.0166,
+      "step": 2219
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4650449639353693e-07,
+      "loss": 0.0213,
+      "step": 2220
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4415671969854272e-07,
+      "loss": 0.0328,
+      "step": 2221
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4182777058754814e-07,
+      "loss": 0.023,
+      "step": 2222
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.395176535095355e-07,
+      "loss": 0.0197,
+      "step": 2223
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.372263728775147e-07,
+      "loss": 0.0336,
+      "step": 2224
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.3495393306851012e-07,
+      "loss": 0.0256,
+      "step": 2225
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.327003384235581e-07,
+      "loss": 0.0275,
+      "step": 2226
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.304655932476906e-07,
+      "loss": 0.0304,
+      "step": 2227
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.282497018099349e-07,
+      "loss": 0.0127,
+      "step": 2228
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2605266834330053e-07,
+      "loss": 0.0213,
+      "step": 2229
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.238744970447714e-07,
+      "loss": 0.034,
+      "step": 2230
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2171519207530457e-07,
+      "loss": 0.0367,
+      "step": 2231
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.195747575598094e-07,
+      "loss": 0.0262,
+      "step": 2232
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.174531975871529e-07,
+      "loss": 0.0205,
+      "step": 2233
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1535051621014315e-07,
+      "loss": 0.0291,
+      "step": 2234
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1326671744552486e-07,
+      "loss": 0.0266,
+      "step": 2235
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1120180527397273e-07,
+      "loss": 0.04,
+      "step": 2236
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0915578364008028e-07,
+      "loss": 0.0261,
+      "step": 2237
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.071286564523566e-07,
+      "loss": 0.0246,
+      "step": 2238
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0512042758321517e-07,
+      "loss": 0.0276,
+      "step": 2239
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0313110086896949e-07,
+      "loss": 0.031,
+      "step": 2240
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0116068010982305e-07,
+      "loss": 0.0334,
+      "step": 2241
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.920916906986377e-08,
+      "loss": 0.0249,
+      "step": 2242
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.727657147705738e-08,
+      "loss": 0.0362,
+      "step": 2243
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.536289102323847e-08,
+      "loss": 0.0367,
+      "step": 2244
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.346813136410393e-08,
+      "loss": 0.0227,
+      "step": 2245
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.159229611920617e-08,
+      "loss": 0.0453,
+      "step": 2246
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.973538887194877e-08,
+      "loss": 0.0277,
+      "step": 2247
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.789741316957312e-08,
+      "loss": 0.0275,
+      "step": 2248
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.607837252315953e-08,
+      "loss": 0.0313,
+      "step": 2249
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.427827040761394e-08,
+      "loss": 0.0247,
+      "step": 2250
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.249711026166673e-08,
+      "loss": 0.0297,
+      "step": 2251
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 8.073489548786062e-08,
+      "loss": 0.0238,
+      "step": 2252
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.899162945254946e-08,
+      "loss": 0.0165,
+      "step": 2253
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.726731548588828e-08,
+      "loss": 0.0233,
+      "step": 2254
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.556195688182887e-08,
+      "loss": 0.0367,
+      "step": 2255
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.387555689810977e-08,
+      "loss": 0.0195,
+      "step": 2256
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.220811875625732e-08,
+      "loss": 0.0336,
+      "step": 2257
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.055964564157247e-08,
+      "loss": 0.0432,
+      "step": 2258
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.893014070312732e-08,
+      "loss": 0.0279,
+      "step": 2259
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.731960705375962e-08,
+      "loss": 0.0199,
+      "step": 2260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.572804777006725e-08,
+      "loss": 0.0197,
+      "step": 2261
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.41554658924004e-08,
+      "loss": 0.0362,
+      "step": 2262
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.260186442485495e-08,
+      "loss": 0.0274,
+      "step": 2263
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.10672463352735e-08,
+      "loss": 0.0337,
+      "step": 2264
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9551614555228845e-08,
+      "loss": 0.0294,
+      "step": 2265
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.805497198002941e-08,
+      "loss": 0.0149,
+      "step": 2266
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.657732146870487e-08,
+      "loss": 0.0206,
+      "step": 2267
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.511866584400838e-08,
+      "loss": 0.0305,
+      "step": 2268
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.36790078924021e-08,
+      "loss": 0.0207,
+      "step": 2269
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.2258350364060574e-08,
+      "loss": 0.0305,
+      "step": 2270
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.0856695972861805e-08,
+      "loss": 0.0227,
+      "step": 2271
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.947404739638062e-08,
+      "loss": 0.0197,
+      "step": 2272
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.8110407275887564e-08,
+      "loss": 0.0184,
+      "step": 2273
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.6765778216336654e-08,
+      "loss": 0.0229,
+      "step": 2274
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.5440162786370976e-08,
+      "loss": 0.0244,
+      "step": 2275
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.413356351830822e-08,
+      "loss": 0.0325,
+      "step": 2276
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.284598290814068e-08,
+      "loss": 0.0223,
+      "step": 2277
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.157742341552862e-08,
+      "loss": 0.0202,
+      "step": 2278
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.032788746379801e-08,
+      "loss": 0.0231,
+      "step": 2279
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.9097377439931695e-08,
+      "loss": 0.0193,
+      "step": 2280
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.788589569457046e-08,
+      "loss": 0.0209,
+      "step": 2281
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.669344454200196e-08,
+      "loss": 0.0298,
+      "step": 2282
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.552002626016293e-08,
+      "loss": 0.0406,
+      "step": 2283
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.436564309063028e-08,
+      "loss": 0.0201,
+      "step": 2284
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.3230297238617816e-08,
+      "loss": 0.0293,
+      "step": 2285
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.2113990872972844e-08,
+      "loss": 0.0274,
+      "step": 2286
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.10167261261729e-08,
+      "loss": 0.0168,
+      "step": 2287
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.993850509431684e-08,
+      "loss": 0.034,
+      "step": 2288
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.8879329837128157e-08,
+      "loss": 0.0216,
+      "step": 2289
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.783920237794502e-08,
+      "loss": 0.0345,
+      "step": 2290
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.681812470372025e-08,
+      "loss": 0.0337,
+      "step": 2291
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.5816098765014673e-08,
+      "loss": 0.0172,
+      "step": 2292
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.4833126475994895e-08,
+      "loss": 0.0339,
+      "step": 2293
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3869209714428855e-08,
+      "loss": 0.0319,
+      "step": 2294
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.2924350321685828e-08,
+      "loss": 0.0254,
+      "step": 2295
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1998550102724225e-08,
+      "loss": 0.039,
+      "step": 2296
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.1091810826099345e-08,
+      "loss": 0.0269,
+      "step": 2297
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.020413422395229e-08,
+      "loss": 0.021,
+      "step": 2298
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9335521992007723e-08,
+      "loss": 0.0217,
+      "step": 2299
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8485975789571674e-08,
+      "loss": 0.034,
+      "step": 2300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.765549723953153e-08,
+      "loss": 0.0212,
+      "step": 2301
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.684408792834713e-08,
+      "loss": 0.0212,
+      "step": 2302
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.6051749406049698e-08,
+      "loss": 0.0207,
+      "step": 2303
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.527848318624181e-08,
+      "loss": 0.0178,
+      "step": 2304
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.4524290746090742e-08,
+      "loss": 0.0284,
+      "step": 2305
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.378917352632958e-08,
+      "loss": 0.0275,
+      "step": 2306
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3073132931247234e-08,
+      "loss": 0.0319,
+      "step": 2307
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2376170328696202e-08,
+      "loss": 0.024,
+      "step": 2308
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1698287050079248e-08,
+      "loss": 0.029,
+      "step": 2309
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.103948439035718e-08,
+      "loss": 0.0241,
+      "step": 2310
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0399763608037738e-08,
+      "loss": 0.0295,
+      "step": 2311
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.779125925175602e-09,
+      "loss": 0.0282,
+      "step": 2312
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.177572527375722e-09,
+      "loss": 0.0232,
+      "step": 2313
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 8.595104563781098e-09,
+      "loss": 0.0301,
+      "step": 2314
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 8.03172314707834e-09,
+      "loss": 0.0356,
+      "step": 2315
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.487429353495445e-09,
+      "loss": 0.0294,
+      "step": 2316
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.962224222792913e-09,
+      "loss": 0.0296,
+      "step": 2317
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.456108758268187e-09,
+      "loss": 0.0182,
+      "step": 2318
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.969083926753439e-09,
+      "loss": 0.0255,
+      "step": 2319
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5011506586100105e-09,
+      "loss": 0.045,
+      "step": 2320
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.0523098477284204e-09,
+      "loss": 0.0221,
+      "step": 2321
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.622562351527249e-09,
+      "loss": 0.0402,
+      "step": 2322
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.21190899095425e-09,
+      "loss": 0.0242,
+      "step": 2323
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.82035055047747e-09,
+      "loss": 0.0231,
+      "step": 2324
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4478877780885765e-09,
+      "loss": 0.0223,
+      "step": 2325
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0945213853039724e-09,
+      "loss": 0.0166,
+      "step": 2326
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.7602520471581297e-09,
+      "loss": 0.0259,
+      "step": 2327
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.4450804022024822e-09,
+      "loss": 0.0262,
+      "step": 2328
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1490070525109763e-09,
+      "loss": 0.0276,
+      "step": 2329
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.872032563670079e-09,
+      "loss": 0.0223,
+      "step": 2330
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.614157464783217e-09,
+      "loss": 0.0149,
+      "step": 2331
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3753822484685596e-09,
+      "loss": 0.0262,
+      "step": 2332
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1557073708579058e-09,
+      "loss": 0.0337,
+      "step": 2333
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.551332515955747e-10,
+      "loss": 0.0274,
+      "step": 2334
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.736602738372956e-10,
+      "loss": 0.0228,
+      "step": 2335
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.112887842490978e-10,
+      "loss": 0.0143,
+      "step": 2336
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.68019093010641e-10,
+      "loss": 0.0351,
+      "step": 2337
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.438514738085541e-10,
+      "loss": 0.0225,
+      "step": 2338
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3878616384087614e-10,
+      "loss": 0.0183,
+      "step": 2339
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.5282336381150508e-10,
+      "loss": 0.0275,
+      "step": 2340
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.59632379357489e-11,
+      "loss": 0.0217,
+      "step": 2341
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.820591393699502e-11,
+      "loss": 0.0301,
+      "step": 2342
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.551483045600052e-12,
+      "loss": 0.0284,
+      "step": 2343
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0,
+      "loss": 0.0236,
+      "step": 2344
+    },
+    {
+      "epoch": 2.0,
+      "step": 2344,
+      "total_flos": 129140992770048.0,
+      "train_loss": 0.06272130045060818,
+      "train_runtime": 7733.9148,
+      "train_samples_per_second": 38.79,
+      "train_steps_per_second": 0.303
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2344,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 11100,
+  "total_flos": 129140992770048.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}