diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.5690430314707773,
+  "epoch": 2.772002772002772,
   "eval_steps": 500,
-  "global_step": 1000,
+  "global_step": 13000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1207,13 +1207,14413 @@
       "learning_rate": 0.0001,
       "loss": 4.1717,
       "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.571337172104927e-05,
+      "loss": 4.4083,
+      "step": 1005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.569204521219876e-05,
+      "loss": 4.1957,
+      "step": 1010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.567071870334826e-05,
+      "loss": 4.2654,
+      "step": 1015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.564939219449776e-05,
+      "loss": 4.251,
+      "step": 1020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.562806568564726e-05,
+      "loss": 4.2639,
+      "step": 1025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.560673917679676e-05,
+      "loss": 4.1973,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.558541266794626e-05,
+      "loss": 4.129,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.556408615909575e-05,
+      "loss": 4.2596,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.554275965024526e-05,
+      "loss": 4.3568,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.552143314139475e-05,
+      "loss": 4.3071,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.550010663254426e-05,
+      "loss": 4.275,
+      "step": 1055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.547878012369376e-05,
+      "loss": 4.1531,
+      "step": 1060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.545745361484326e-05,
+      "loss": 4.2054,
+      "step": 1065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.543612710599276e-05,
+      "loss": 4.3501,
+      "step": 1070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.541480059714226e-05,
+      "loss": 4.2209,
+      "step": 1075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.539347408829176e-05,
+      "loss": 4.3029,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.537214757944126e-05,
+      "loss": 4.2065,
+      "step": 1085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.535082107059075e-05,
+      "loss": 4.0332,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.532949456174026e-05,
+      "loss": 4.2867,
+      "step": 1095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.530816805288975e-05,
+      "loss": 4.2094,
+      "step": 1100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.528684154403924e-05,
+      "loss": 4.2344,
+      "step": 1105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.526551503518875e-05,
+      "loss": 4.3236,
+      "step": 1110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.524418852633824e-05,
+      "loss": 4.3949,
+      "step": 1115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.522286201748775e-05,
+      "loss": 4.2918,
+      "step": 1120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.520153550863724e-05,
+      "loss": 4.1913,
+      "step": 1125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.518020899978674e-05,
+      "loss": 4.2062,
+      "step": 1130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.515888249093624e-05,
+      "loss": 4.2299,
+      "step": 1135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.513755598208574e-05,
+      "loss": 4.3388,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.511622947323524e-05,
+      "loss": 4.1539,
+      "step": 1145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.509490296438474e-05,
+      "loss": 4.2461,
+      "step": 1150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.507357645553423e-05,
+      "loss": 4.2264,
+      "step": 1155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.505224994668374e-05,
+      "loss": 4.1781,
+      "step": 1160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.503092343783323e-05,
+      "loss": 4.2005,
+      "step": 1165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.500959692898272e-05,
+      "loss": 4.2533,
+      "step": 1170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.498827042013223e-05,
+      "loss": 4.1824,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.496694391128172e-05,
+      "loss": 4.177,
+      "step": 1180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.494561740243123e-05,
+      "loss": 4.24,
+      "step": 1185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.492429089358072e-05,
+      "loss": 4.2081,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.490296438473022e-05,
+      "loss": 4.3068,
+      "step": 1195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.488163787587972e-05,
+      "loss": 4.2077,
+      "step": 1200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.486031136702922e-05,
+      "loss": 4.2133,
+      "step": 1205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.483898485817872e-05,
+      "loss": 4.2082,
+      "step": 1210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.481765834932822e-05,
+      "loss": 4.1653,
+      "step": 1215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.479633184047771e-05,
+      "loss": 4.2722,
+      "step": 1220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.477500533162722e-05,
+      "loss": 4.2118,
+      "step": 1225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.475367882277671e-05,
+      "loss": 4.2188,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.473235231392622e-05,
+      "loss": 4.2077,
+      "step": 1235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.471102580507571e-05,
+      "loss": 4.2659,
+      "step": 1240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.468969929622522e-05,
+      "loss": 4.2111,
+      "step": 1245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.466837278737471e-05,
+      "loss": 4.0976,
+      "step": 1250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.464704627852422e-05,
+      "loss": 4.1634,
+      "step": 1255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.462571976967371e-05,
+      "loss": 4.219,
+      "step": 1260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.460439326082322e-05,
+      "loss": 4.1829,
+      "step": 1265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.458306675197271e-05,
+      "loss": 4.2319,
+      "step": 1270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.456174024312222e-05,
+      "loss": 4.1322,
+      "step": 1275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.454041373427171e-05,
+      "loss": 4.1847,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.45190872254212e-05,
+      "loss": 4.1901,
+      "step": 1285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.449776071657071e-05,
+      "loss": 4.2486,
+      "step": 1290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.44764342077202e-05,
+      "loss": 4.2425,
+      "step": 1295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.445510769886971e-05,
+      "loss": 4.2159,
+      "step": 1300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.44337811900192e-05,
+      "loss": 4.3078,
+      "step": 1305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.44124546811687e-05,
+      "loss": 4.2446,
+      "step": 1310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.43911281723182e-05,
+      "loss": 4.1374,
+      "step": 1315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.43698016634677e-05,
+      "loss": 4.1335,
+      "step": 1320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.434847515461719e-05,
+      "loss": 4.207,
+      "step": 1325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.43271486457667e-05,
+      "loss": 4.2393,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.430582213691619e-05,
+      "loss": 4.2811,
+      "step": 1335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.42844956280657e-05,
+      "loss": 4.117,
+      "step": 1340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.426316911921519e-05,
+      "loss": 4.219,
+      "step": 1345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.424184261036468e-05,
+      "loss": 4.1704,
+      "step": 1350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.422051610151419e-05,
+      "loss": 4.1236,
+      "step": 1355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.419918959266368e-05,
+      "loss": 4.1909,
+      "step": 1360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.417786308381319e-05,
+      "loss": 4.1695,
+      "step": 1365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.415653657496268e-05,
+      "loss": 4.2374,
+      "step": 1370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.413521006611218e-05,
+      "loss": 4.1597,
+      "step": 1375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.411388355726168e-05,
+      "loss": 4.2953,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.409255704841118e-05,
+      "loss": 4.2969,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.407123053956067e-05,
+      "loss": 4.2616,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.404990403071018e-05,
+      "loss": 4.1079,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.402857752185967e-05,
+      "loss": 4.1583,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.400725101300918e-05,
+      "loss": 4.2673,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.398592450415867e-05,
+      "loss": 4.0916,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.396459799530816e-05,
+      "loss": 4.2036,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.394327148645767e-05,
+      "loss": 4.1299,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.392194497760716e-05,
+      "loss": 4.1743,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.390061846875667e-05,
+      "loss": 4.1222,
+      "step": 1430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.387929195990616e-05,
+      "loss": 4.2357,
+      "step": 1435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.385796545105567e-05,
+      "loss": 4.1811,
+      "step": 1440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.383663894220516e-05,
+      "loss": 4.2133,
+      "step": 1445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.381531243335467e-05,
+      "loss": 4.1,
+      "step": 1450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.379398592450416e-05,
+      "loss": 4.1854,
+      "step": 1455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.377265941565367e-05,
+      "loss": 4.2457,
+      "step": 1460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.375133290680316e-05,
+      "loss": 4.089,
+      "step": 1465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.373000639795267e-05,
+      "loss": 4.1352,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.370867988910216e-05,
+      "loss": 4.1776,
+      "step": 1475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.368735338025166e-05,
+      "loss": 4.0168,
+      "step": 1480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.366602687140116e-05,
+      "loss": 4.2118,
+      "step": 1485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.364470036255066e-05,
+      "loss": 4.2413,
+      "step": 1490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.362337385370016e-05,
+      "loss": 4.1838,
+      "step": 1495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.360204734484966e-05,
+      "loss": 4.1858,
+      "step": 1500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.358072083599915e-05,
+      "loss": 4.1478,
+      "step": 1505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.355939432714866e-05,
+      "loss": 4.2252,
+      "step": 1510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.353806781829815e-05,
+      "loss": 4.1753,
+      "step": 1515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.351674130944766e-05,
+      "loss": 4.1992,
+      "step": 1520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.349541480059715e-05,
+      "loss": 4.0956,
+      "step": 1525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.347408829174664e-05,
+      "loss": 4.1486,
+      "step": 1530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.345276178289615e-05,
+      "loss": 4.113,
+      "step": 1535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.343143527404564e-05,
+      "loss": 4.1695,
+      "step": 1540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.341010876519514e-05,
+      "loss": 4.1399,
+      "step": 1545
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.338878225634464e-05,
+      "loss": 4.1361,
+      "step": 1550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.336745574749414e-05,
+      "loss": 4.1291,
+      "step": 1555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.334612923864364e-05,
+      "loss": 4.1729,
+      "step": 1560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.332480272979314e-05,
+      "loss": 4.186,
+      "step": 1565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.330347622094263e-05,
+      "loss": 4.1694,
+      "step": 1570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.328214971209214e-05,
+      "loss": 4.2573,
+      "step": 1575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.326082320324163e-05,
+      "loss": 4.1644,
+      "step": 1580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.323949669439114e-05,
+      "loss": 4.1274,
+      "step": 1585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.321817018554063e-05,
+      "loss": 4.2186,
+      "step": 1590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.319684367669012e-05,
+      "loss": 4.1449,
+      "step": 1595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.317551716783963e-05,
+      "loss": 4.2566,
+      "step": 1600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.315419065898912e-05,
+      "loss": 4.1832,
+      "step": 1605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.313286415013863e-05,
+      "loss": 4.2554,
+      "step": 1610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.311153764128812e-05,
+      "loss": 4.2106,
+      "step": 1615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.309021113243761e-05,
+      "loss": 4.2412,
+      "step": 1620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.306888462358712e-05,
+      "loss": 4.2214,
+      "step": 1625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.304755811473661e-05,
+      "loss": 4.124,
+      "step": 1630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.302623160588612e-05,
+      "loss": 4.2064,
+      "step": 1635
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.300490509703563e-05,
+      "loss": 4.2735,
+      "step": 1640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.298357858818512e-05,
+      "loss": 4.1409,
+      "step": 1645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.296225207933463e-05,
+      "loss": 4.1364,
+      "step": 1650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.294092557048412e-05,
+      "loss": 4.1741,
+      "step": 1655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.291959906163361e-05,
+      "loss": 4.2158,
+      "step": 1660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.289827255278312e-05,
+      "loss": 4.1697,
+      "step": 1665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.287694604393261e-05,
+      "loss": 4.1616,
+      "step": 1670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.285561953508212e-05,
+      "loss": 4.0912,
+      "step": 1675
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.283429302623161e-05,
+      "loss": 4.1208,
+      "step": 1680
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.281296651738111e-05,
+      "loss": 4.2515,
+      "step": 1685
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.279164000853062e-05,
+      "loss": 4.2275,
+      "step": 1690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.277031349968011e-05,
+      "loss": 4.1442,
+      "step": 1695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.27489869908296e-05,
+      "loss": 4.3005,
+      "step": 1700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.272766048197911e-05,
+      "loss": 4.1924,
+      "step": 1705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.27063339731286e-05,
+      "loss": 4.2763,
+      "step": 1710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.268500746427811e-05,
+      "loss": 4.1912,
+      "step": 1715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.26636809554276e-05,
+      "loss": 4.1089,
+      "step": 1720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.26423544465771e-05,
+      "loss": 4.1599,
+      "step": 1725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.26210279377266e-05,
+      "loss": 4.2416,
+      "step": 1730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.25997014288761e-05,
+      "loss": 4.2332,
+      "step": 1735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.25783749200256e-05,
+      "loss": 4.1482,
+      "step": 1740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.25570484111751e-05,
+      "loss": 4.2588,
+      "step": 1745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.253572190232459e-05,
+      "loss": 4.2557,
+      "step": 1750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.25143953934741e-05,
+      "loss": 4.2302,
+      "step": 1755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.249306888462359e-05,
+      "loss": 4.1986,
+      "step": 1760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.247174237577308e-05,
+      "loss": 4.1768,
+      "step": 1765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.245041586692259e-05,
+      "loss": 4.2593,
+      "step": 1770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.242908935807208e-05,
+      "loss": 4.0778,
+      "step": 1775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.240776284922159e-05,
+      "loss": 4.1143,
+      "step": 1780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.238643634037108e-05,
+      "loss": 4.2432,
+      "step": 1785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.236510983152057e-05,
+      "loss": 4.1342,
+      "step": 1790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.234378332267008e-05,
+      "loss": 4.1526,
+      "step": 1795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.232245681381957e-05,
+      "loss": 4.1949,
+      "step": 1800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.230113030496908e-05,
+      "loss": 4.1611,
+      "step": 1805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.227980379611857e-05,
+      "loss": 4.2797,
+      "step": 1810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.225847728726808e-05,
+      "loss": 4.1547,
+      "step": 1815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.223715077841757e-05,
+      "loss": 4.21,
+      "step": 1820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.221582426956708e-05,
+      "loss": 4.1497,
+      "step": 1825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.219449776071657e-05,
+      "loss": 4.2368,
+      "step": 1830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.217317125186608e-05,
+      "loss": 4.1244,
+      "step": 1835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.215184474301557e-05,
+      "loss": 4.1749,
+      "step": 1840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.213051823416508e-05,
+      "loss": 4.1415,
+      "step": 1845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.210919172531457e-05,
+      "loss": 4.1345,
+      "step": 1850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.208786521646407e-05,
+      "loss": 4.109,
+      "step": 1855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.206653870761357e-05,
+      "loss": 4.2461,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.204521219876307e-05,
+      "loss": 4.2095,
+      "step": 1865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.202388568991257e-05,
+      "loss": 4.0995,
+      "step": 1870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.200255918106207e-05,
+      "loss": 4.1551,
+      "step": 1875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.198123267221156e-05,
+      "loss": 4.1101,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.195990616336107e-05,
+      "loss": 4.1769,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.193857965451056e-05,
+      "loss": 4.1166,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.191725314566007e-05,
+      "loss": 4.1728,
+      "step": 1895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.189592663680956e-05,
+      "loss": 4.1185,
+      "step": 1900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.187460012795905e-05,
+      "loss": 4.1217,
+      "step": 1905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.185327361910856e-05,
+      "loss": 4.2622,
+      "step": 1910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.183194711025805e-05,
+      "loss": 4.1488,
+      "step": 1915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.181062060140755e-05,
+      "loss": 4.1439,
+      "step": 1920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.178929409255705e-05,
+      "loss": 4.0916,
+      "step": 1925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.176796758370655e-05,
+      "loss": 4.2217,
+      "step": 1930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.174664107485605e-05,
+      "loss": 4.1477,
+      "step": 1935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.172531456600555e-05,
+      "loss": 4.1755,
+      "step": 1940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.170398805715504e-05,
+      "loss": 4.2345,
+      "step": 1945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.168266154830455e-05,
+      "loss": 4.219,
+      "step": 1950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.166133503945404e-05,
+      "loss": 4.1584,
+      "step": 1955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.164000853060355e-05,
+      "loss": 4.2019,
+      "step": 1960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.161868202175304e-05,
+      "loss": 4.2094,
+      "step": 1965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.159735551290253e-05,
+      "loss": 4.0424,
+      "step": 1970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.157602900405204e-05,
+      "loss": 4.1501,
+      "step": 1975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.155470249520153e-05,
+      "loss": 4.1649,
+      "step": 1980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.153337598635103e-05,
+      "loss": 4.1051,
+      "step": 1985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.151204947750053e-05,
+      "loss": 4.188,
+      "step": 1990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.149072296865003e-05,
+      "loss": 4.212,
+      "step": 1995
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.146939645979953e-05,
+      "loss": 4.1834,
+      "step": 2000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.91448069950949e-05,
+      "loss": 4.1643,
+      "step": 2005
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.914267434420986e-05,
+      "loss": 4.0988,
+      "step": 2010
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.91405416933248e-05,
+      "loss": 4.1608,
+      "step": 2015
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.913840904243976e-05,
+      "loss": 4.1784,
+      "step": 2020
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.913627639155471e-05,
+      "loss": 4.0438,
+      "step": 2025
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.913414374066967e-05,
+      "loss": 4.2819,
+      "step": 2030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.913201108978461e-05,
+      "loss": 4.1087,
+      "step": 2035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.912987843889956e-05,
+      "loss": 4.137,
+      "step": 2040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.91277457880145e-05,
+      "loss": 4.176,
+      "step": 2045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.912561313712946e-05,
+      "loss": 4.1253,
+      "step": 2050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.91234804862444e-05,
+      "loss": 4.1579,
+      "step": 2055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.912134783535935e-05,
+      "loss": 4.1829,
+      "step": 2060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.91192151844743e-05,
+      "loss": 4.2767,
+      "step": 2065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.911708253358925e-05,
+      "loss": 4.1685,
+      "step": 2070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.91149498827042e-05,
+      "loss": 4.1432,
+      "step": 2075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.911281723181916e-05,
+      "loss": 4.0659,
+      "step": 2080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.911068458093411e-05,
+      "loss": 4.1739,
+      "step": 2085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.910855193004906e-05,
+      "loss": 4.1117,
+      "step": 2090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.910641927916401e-05,
+      "loss": 4.2319,
+      "step": 2095
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.910428662827895e-05,
+      "loss": 4.2685,
+      "step": 2100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.91021539773939e-05,
+      "loss": 4.1428,
+      "step": 2105
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.910002132650885e-05,
+      "loss": 4.1356,
+      "step": 2110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.90978886756238e-05,
+      "loss": 4.1466,
+      "step": 2115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.909575602473876e-05,
+      "loss": 4.1401,
+      "step": 2120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.909362337385371e-05,
+      "loss": 4.1408,
+      "step": 2125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.909149072296865e-05,
+      "loss": 4.0734,
+      "step": 2130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.908935807208361e-05,
+      "loss": 4.1361,
+      "step": 2135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.908722542119856e-05,
+      "loss": 4.036,
+      "step": 2140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.90850927703135e-05,
+      "loss": 4.1354,
+      "step": 2145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.908296011942846e-05,
+      "loss": 4.0682,
+      "step": 2150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.90808274685434e-05,
+      "loss": 4.1486,
+      "step": 2155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.907869481765835e-05,
+      "loss": 4.1871,
+      "step": 2160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.90765621667733e-05,
+      "loss": 4.1501,
+      "step": 2165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.907442951588825e-05,
+      "loss": 4.1746,
+      "step": 2170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.90722968650032e-05,
+      "loss": 4.0955,
+      "step": 2175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.907016421411816e-05,
+      "loss": 4.1039,
+      "step": 2180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.90680315632331e-05,
+      "loss": 4.2038,
+      "step": 2185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.906589891234806e-05,
+      "loss": 4.1666,
+      "step": 2190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.9063766261463e-05,
+      "loss": 4.1018,
+      "step": 2195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.906163361057795e-05,
+      "loss": 4.2162,
+      "step": 2200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.90595009596929e-05,
+      "loss": 4.0588,
+      "step": 2205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.905736830880785e-05,
+      "loss": 4.0441,
+      "step": 2210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.90552356579228e-05,
+      "loss": 4.1408,
+      "step": 2215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.905310300703776e-05,
+      "loss": 4.1757,
+      "step": 2220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.905097035615271e-05,
+      "loss": 4.2243,
+      "step": 2225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.904883770526765e-05,
+      "loss": 4.1703,
+      "step": 2230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.904670505438261e-05,
+      "loss": 4.0536,
+      "step": 2235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.904457240349755e-05,
+      "loss": 4.1506,
+      "step": 2240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.90424397526125e-05,
+      "loss": 4.2364,
+      "step": 2245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.904030710172744e-05,
+      "loss": 4.2115,
+      "step": 2250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.90381744508424e-05,
+      "loss": 4.0522,
+      "step": 2255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.903604179995734e-05,
+      "loss": 4.2237,
+      "step": 2260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.90339091490723e-05,
+      "loss": 4.1049,
+      "step": 2265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.903177649818726e-05,
+      "loss": 4.1546,
+      "step": 2270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.90296438473022e-05,
+      "loss": 4.0804,
+      "step": 2275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.902751119641716e-05,
+      "loss": 4.132,
+      "step": 2280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.90253785455321e-05,
+      "loss": 4.093,
+      "step": 2285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.902324589464706e-05,
+      "loss": 4.154,
+      "step": 2290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.9021113243762e-05,
+      "loss": 4.1667,
+      "step": 2295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.901898059287695e-05,
+      "loss": 4.0874,
+      "step": 2300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.901684794199189e-05,
+      "loss": 4.1243,
+      "step": 2305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.901471529110685e-05,
+      "loss": 4.2504,
+      "step": 2310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.90125826402218e-05,
+      "loss": 4.0576,
+      "step": 2315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.901044998933676e-05,
+      "loss": 4.1209,
+      "step": 2320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.90083173384517e-05,
+      "loss": 4.1222,
+      "step": 2325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.900618468756665e-05,
+      "loss": 4.2048,
+      "step": 2330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.900405203668161e-05,
+      "loss": 4.1303,
+      "step": 2335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.900191938579655e-05,
+      "loss": 4.0732,
+      "step": 2340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.89997867349115e-05,
+      "loss": 4.1394,
+      "step": 2345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.899765408402644e-05,
+      "loss": 3.9982,
+      "step": 2350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.89955214331414e-05,
+      "loss": 4.0899,
+      "step": 2355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.899338878225634e-05,
+      "loss": 4.2053,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.89912561313713e-05,
+      "loss": 4.1326,
+      "step": 2365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.898912348048625e-05,
+      "loss": 4.11,
+      "step": 2370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.89869908296012e-05,
+      "loss": 4.1633,
+      "step": 2375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.898485817871615e-05,
+      "loss": 4.1121,
+      "step": 2380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.89827255278311e-05,
+      "loss": 4.0735,
+      "step": 2385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.898059287694604e-05,
+      "loss": 4.1586,
+      "step": 2390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.8978460226061e-05,
+      "loss": 4.2001,
+      "step": 2395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.897632757517595e-05,
+      "loss": 4.1727,
+      "step": 2400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.897419492429089e-05,
+      "loss": 4.1287,
+      "step": 2405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.897206227340585e-05,
+      "loss": 4.212,
+      "step": 2410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.89699296225208e-05,
+      "loss": 4.1361,
+      "step": 2415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896779697163576e-05,
+      "loss": 4.0663,
+      "step": 2420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89656643207507e-05,
+      "loss": 4.1787,
+      "step": 2425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896353166986565e-05,
+      "loss": 4.1471,
+      "step": 2430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89613990189806e-05,
+      "loss": 4.1378,
+      "step": 2435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895926636809555e-05,
+      "loss": 4.1138,
+      "step": 2440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895713371721049e-05,
+      "loss": 4.1356,
+      "step": 2445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895500106632544e-05,
+      "loss": 4.1292,
+      "step": 2450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895286841544039e-05,
+      "loss": 4.2884,
+      "step": 2455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895073576455534e-05,
+      "loss": 4.1709,
+      "step": 2460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.894860311367031e-05,
+      "loss": 4.1509,
+      "step": 2465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.894647046278525e-05,
+      "loss": 4.1469,
+      "step": 2470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.89443378119002e-05,
+      "loss": 4.1039,
+      "step": 2475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.894220516101515e-05,
+      "loss": 4.0355,
+      "step": 2480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.89400725101301e-05,
+      "loss": 4.1512,
+      "step": 2485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.893793985924504e-05,
+      "loss": 4.2215,
+      "step": 2490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.893580720836e-05,
+      "loss": 4.0549,
+      "step": 2495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.893367455747494e-05,
+      "loss": 4.0303,
+      "step": 2500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.893154190658989e-05,
+      "loss": 4.0719,
+      "step": 2505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.892940925570485e-05,
+      "loss": 4.1445,
+      "step": 2510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.89272766048198e-05,
+      "loss": 4.0641,
+      "step": 2515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.892514395393474e-05,
+      "loss": 4.124,
+      "step": 2520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.89230113030497e-05,
+      "loss": 4.1024,
+      "step": 2525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.892087865216465e-05,
+      "loss": 4.0906,
+      "step": 2530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.89187460012796e-05,
+      "loss": 4.1771,
+      "step": 2535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.891661335039455e-05,
+      "loss": 4.1613,
+      "step": 2540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.891448069950949e-05,
+      "loss": 4.052,
+      "step": 2545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.891234804862444e-05,
+      "loss": 4.0764,
+      "step": 2550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.891021539773939e-05,
+      "loss": 4.1504,
+      "step": 2555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890808274685434e-05,
+      "loss": 4.0837,
+      "step": 2560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.89059500959693e-05,
+      "loss": 4.0608,
+      "step": 2565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890381744508425e-05,
+      "loss": 4.0945,
+      "step": 2570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890168479419919e-05,
+      "loss": 4.1251,
+      "step": 2575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889955214331415e-05,
+      "loss": 4.1016,
+      "step": 2580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889741949242909e-05,
+      "loss": 4.0497,
+      "step": 2585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889528684154404e-05,
+      "loss": 4.0531,
+      "step": 2590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.8893154190659e-05,
+      "loss": 4.1791,
+      "step": 2595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889102153977394e-05,
+      "loss": 4.0953,
+      "step": 2600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.888888888888889e-05,
+      "loss": 4.1382,
+      "step": 2605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.888675623800385e-05,
+      "loss": 4.1342,
+      "step": 2610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.88846235871188e-05,
+      "loss": 4.1683,
+      "step": 2615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.888249093623374e-05,
+      "loss": 4.1578,
+      "step": 2620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.88803582853487e-05,
+      "loss": 4.1064,
+      "step": 2625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.887822563446364e-05,
+      "loss": 4.1793,
+      "step": 2630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.88760929835786e-05,
+      "loss": 4.0855,
+      "step": 2635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.887396033269353e-05,
+      "loss": 4.001,
+      "step": 2640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.887182768180849e-05,
+      "loss": 4.1257,
+      "step": 2645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.886969503092343e-05,
+      "loss": 4.0758,
+      "step": 2650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.886756238003839e-05,
+      "loss": 4.1618,
+      "step": 2655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.886542972915335e-05,
+      "loss": 3.9996,
+      "step": 2660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.88632970782683e-05,
+      "loss": 3.9529,
+      "step": 2665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.886116442738325e-05,
+      "loss": 4.0993,
+      "step": 2670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.885903177649819e-05,
+      "loss": 4.1167,
+      "step": 2675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.885689912561315e-05,
+      "loss": 4.0109,
+      "step": 2680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.885476647472809e-05,
+      "loss": 4.1242,
+      "step": 2685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.885263382384304e-05,
+      "loss": 4.1102,
+      "step": 2690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.885050117295798e-05,
+      "loss": 4.1421,
+      "step": 2695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884836852207294e-05,
+      "loss": 4.1446,
+      "step": 2700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884623587118789e-05,
+      "loss": 4.0766,
+      "step": 2705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884410322030285e-05,
+      "loss": 4.2073,
+      "step": 2710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884197056941779e-05,
+      "loss": 4.0118,
+      "step": 2715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.883983791853274e-05,
+      "loss": 4.052,
+      "step": 2720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.88377052676477e-05,
+      "loss": 4.2132,
+      "step": 2725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.883557261676264e-05,
+      "loss": 4.0909,
+      "step": 2730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.88334399658776e-05,
+      "loss": 4.0882,
+      "step": 2735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.883130731499253e-05,
+      "loss": 4.0401,
+      "step": 2740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.882917466410749e-05,
+      "loss": 4.046,
+      "step": 2745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.882704201322243e-05,
+      "loss": 4.0792,
+      "step": 2750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.882490936233739e-05,
+      "loss": 4.1257,
+      "step": 2755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.882277671145234e-05,
+      "loss": 4.0106,
+      "step": 2760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.88206440605673e-05,
+      "loss": 4.0985,
+      "step": 2765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.881851140968224e-05,
+      "loss": 4.0917,
+      "step": 2770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.881637875879719e-05,
+      "loss": 4.0979,
+      "step": 2775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.881424610791213e-05,
+      "loss": 4.0565,
+      "step": 2780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.881211345702709e-05,
+      "loss": 4.1683,
+      "step": 2785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.880998080614204e-05,
+      "loss": 4.143,
+      "step": 2790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.880784815525698e-05,
+      "loss": 4.034,
+      "step": 2795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.880571550437194e-05,
+      "loss": 4.1128,
+      "step": 2800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.880358285348689e-05,
+      "loss": 4.1253,
+      "step": 2805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.880145020260185e-05,
+      "loss": 4.0372,
+      "step": 2810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.879931755171679e-05,
+      "loss": 4.1651,
+      "step": 2815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.879718490083174e-05,
+      "loss": 4.1341,
+      "step": 2820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.879505224994668e-05,
+      "loss": 4.0892,
+      "step": 2825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.879291959906164e-05,
+      "loss": 4.1236,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.879078694817658e-05,
+      "loss": 4.1719,
+      "step": 2835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.878865429729153e-05,
+      "loss": 4.1386,
+      "step": 2840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.878652164640649e-05,
+      "loss": 4.1138,
+      "step": 2845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.878438899552143e-05,
+      "loss": 4.1469,
+      "step": 2850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.87822563446364e-05,
+      "loss": 4.1446,
+      "step": 2855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.878012369375134e-05,
+      "loss": 4.1322,
+      "step": 2860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.87779910428663e-05,
+      "loss": 4.0845,
+      "step": 2865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.877585839198124e-05,
+      "loss": 3.9765,
+      "step": 2870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.877372574109619e-05,
+      "loss": 4.011,
+      "step": 2875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.877159309021113e-05,
+      "loss": 4.1012,
+      "step": 2880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.876946043932609e-05,
+      "loss": 4.0735,
+      "step": 2885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.876732778844103e-05,
+      "loss": 4.1378,
+      "step": 2890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.876519513755598e-05,
+      "loss": 4.0898,
+      "step": 2895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.876306248667094e-05,
+      "loss": 3.9766,
+      "step": 2900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.876092983578589e-05,
+      "loss": 4.0375,
+      "step": 2905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875879718490085e-05,
+      "loss": 4.1669,
+      "step": 2910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875666453401579e-05,
+      "loss": 4.1662,
+      "step": 2915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875453188313074e-05,
+      "loss": 4.032,
+      "step": 2920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875239923224568e-05,
+      "loss": 4.1211,
+      "step": 2925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875026658136064e-05,
+      "loss": 4.0966,
+      "step": 2930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.874813393047558e-05,
+      "loss": 4.0909,
+      "step": 2935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.874600127959053e-05,
+      "loss": 4.1091,
+      "step": 2940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.874386862870548e-05,
+      "loss": 4.0977,
+      "step": 2945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.874173597782044e-05,
+      "loss": 4.0881,
+      "step": 2950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.873960332693539e-05,
+      "loss": 4.1552,
+      "step": 2955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.873747067605034e-05,
+      "loss": 4.0706,
+      "step": 2960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.873533802516528e-05,
+      "loss": 4.0482,
+      "step": 2965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.873320537428024e-05,
+      "loss": 4.1118,
+      "step": 2970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.873107272339519e-05,
+      "loss": 4.1406,
+      "step": 2975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.872894007251013e-05,
+      "loss": 4.0668,
+      "step": 2980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.872680742162509e-05,
+      "loss": 4.1856,
+      "step": 2985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.872467477074003e-05,
+      "loss": 4.1681,
+      "step": 2990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.872254211985498e-05,
+      "loss": 4.1388,
+      "step": 2995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.872040946896994e-05,
+      "loss": 4.0494,
+      "step": 3000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.871827681808489e-05,
+      "loss": 4.1185,
+      "step": 3005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.871614416719983e-05,
+      "loss": 4.1959,
+      "step": 3010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.871401151631479e-05,
+      "loss": 4.1379,
+      "step": 3015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.871187886542973e-05,
+      "loss": 4.1076,
+      "step": 3020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870974621454468e-05,
+      "loss": 4.1256,
+      "step": 3025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870761356365963e-05,
+      "loss": 4.0688,
+      "step": 3030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870548091277458e-05,
+      "loss": 4.1218,
+      "step": 3035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870334826188953e-05,
+      "loss": 4.061,
+      "step": 3040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870121561100448e-05,
+      "loss": 4.05,
+      "step": 3045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.869908296011944e-05,
+      "loss": 4.0169,
+      "step": 3050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.869695030923439e-05,
+      "loss": 4.127,
+      "step": 3055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.869481765834934e-05,
+      "loss": 3.9995,
+      "step": 3060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.869268500746428e-05,
+      "loss": 4.1399,
+      "step": 3065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.869055235657924e-05,
+      "loss": 4.1214,
+      "step": 3070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.868841970569418e-05,
+      "loss": 4.1678,
+      "step": 3075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.868628705480913e-05,
+      "loss": 4.0547,
+      "step": 3080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.868415440392407e-05,
+      "loss": 4.1074,
+      "step": 3085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.868202175303903e-05,
+      "loss": 4.1286,
+      "step": 3090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.867988910215398e-05,
+      "loss": 4.1574,
+      "step": 3095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.867775645126894e-05,
+      "loss": 4.1979,
+      "step": 3100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.867562380038389e-05,
+      "loss": 4.0781,
+      "step": 3105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.867349114949883e-05,
+      "loss": 4.1659,
+      "step": 3110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.867135849861379e-05,
+      "loss": 4.0674,
+      "step": 3115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.866922584772873e-05,
+      "loss": 4.1713,
+      "step": 3120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.866709319684368e-05,
+      "loss": 4.1347,
+      "step": 3125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.866496054595863e-05,
+      "loss": 4.0761,
+      "step": 3130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.866282789507358e-05,
+      "loss": 4.0691,
+      "step": 3135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.866069524418852e-05,
+      "loss": 4.0705,
+      "step": 3140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.865856259330349e-05,
+      "loss": 4.0338,
+      "step": 3145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.865642994241843e-05,
+      "loss": 4.1374,
+      "step": 3150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.865429729153339e-05,
+      "loss": 4.0365,
+      "step": 3155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.865216464064833e-05,
+      "loss": 4.0976,
+      "step": 3160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.865003198976328e-05,
+      "loss": 4.0604,
+      "step": 3165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864789933887824e-05,
+      "loss": 4.1707,
+      "step": 3170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864576668799318e-05,
+      "loss": 4.0944,
+      "step": 3175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864363403710813e-05,
+      "loss": 4.0963,
+      "step": 3180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864150138622307e-05,
+      "loss": 4.0062,
+      "step": 3185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863936873533803e-05,
+      "loss": 3.9077,
+      "step": 3190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863723608445298e-05,
+      "loss": 4.0919,
+      "step": 3195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863510343356794e-05,
+      "loss": 4.1093,
+      "step": 3200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863297078268288e-05,
+      "loss": 4.0909,
+      "step": 3205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863083813179783e-05,
+      "loss": 4.1038,
+      "step": 3210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.862870548091277e-05,
+      "loss": 4.042,
+      "step": 3215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.862657283002773e-05,
+      "loss": 4.0529,
+      "step": 3220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.862444017914267e-05,
+      "loss": 4.013,
+      "step": 3225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.862230752825763e-05,
+      "loss": 4.0706,
+      "step": 3230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.862017487737258e-05,
+      "loss": 4.0464,
+      "step": 3235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.861804222648752e-05,
+      "loss": 4.11,
+      "step": 3240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.861590957560249e-05,
+      "loss": 4.0446,
+      "step": 3245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.861377692471743e-05,
+      "loss": 4.2276,
+      "step": 3250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.861164427383239e-05,
+      "loss": 3.9942,
+      "step": 3255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.860951162294733e-05,
+      "loss": 4.094,
+      "step": 3260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.860737897206228e-05,
+      "loss": 4.0907,
+      "step": 3265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.860524632117722e-05,
+      "loss": 4.0857,
+      "step": 3270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.860311367029218e-05,
+      "loss": 4.0191,
+      "step": 3275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.860098101940712e-05,
+      "loss": 4.0758,
+      "step": 3280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.859884836852207e-05,
+      "loss": 4.1778,
+      "step": 3285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.859671571763703e-05,
+      "loss": 4.2016,
+      "step": 3290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.859458306675198e-05,
+      "loss": 3.9973,
+      "step": 3295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.859245041586694e-05,
+      "loss": 4.0544,
+      "step": 3300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.859031776498188e-05,
+      "loss": 4.0442,
+      "step": 3305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.858818511409683e-05,
+      "loss": 4.1261,
+      "step": 3310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.858605246321177e-05,
+      "loss": 4.2476,
+      "step": 3315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.858391981232673e-05,
+      "loss": 3.9749,
+      "step": 3320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.858178716144167e-05,
+      "loss": 4.1603,
+      "step": 3325
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857965451055663e-05,
+      "loss": 4.1023,
+      "step": 3330
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857752185967157e-05,
+      "loss": 4.0706,
+      "step": 3335
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857538920878653e-05,
+      "loss": 4.1547,
+      "step": 3340
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857325655790148e-05,
+      "loss": 4.1731,
+      "step": 3345
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857112390701643e-05,
+      "loss": 3.9327,
+      "step": 3350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.856899125613137e-05,
+      "loss": 4.099,
+      "step": 3355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.856685860524633e-05,
+      "loss": 4.0212,
+      "step": 3360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.856472595436128e-05,
+      "loss": 4.0144,
+      "step": 3365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.856259330347622e-05,
+      "loss": 4.1351,
+      "step": 3370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.856046065259118e-05,
+      "loss": 4.0463,
+      "step": 3375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855832800170612e-05,
+      "loss": 4.1414,
+      "step": 3380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855619535082107e-05,
+      "loss": 4.0608,
+      "step": 3385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855406269993603e-05,
+      "loss": 3.9693,
+      "step": 3390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855193004905098e-05,
+      "loss": 4.1074,
+      "step": 3395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.854979739816592e-05,
+      "loss": 4.106,
+      "step": 3400
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.854766474728088e-05,
+      "loss": 4.0068,
+      "step": 3405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.854553209639582e-05,
+      "loss": 4.0362,
+      "step": 3410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.854339944551077e-05,
+      "loss": 4.0621,
+      "step": 3415
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.854126679462572e-05,
+      "loss": 4.0533,
+      "step": 3420
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853913414374067e-05,
+      "loss": 4.1254,
+      "step": 3425
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853700149285563e-05,
+      "loss": 4.1176,
+      "step": 3430
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853486884197057e-05,
+      "loss": 4.1032,
+      "step": 3435
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853273619108553e-05,
+      "loss": 4.0994,
+      "step": 3440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853060354020048e-05,
+      "loss": 4.0515,
+      "step": 3445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.852847088931543e-05,
+      "loss": 4.0812,
+      "step": 3450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.852633823843037e-05,
+      "loss": 4.0606,
+      "step": 3455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.852420558754533e-05,
+      "loss": 4.0684,
+      "step": 3460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.852207293666027e-05,
+      "loss": 4.0667,
+      "step": 3465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.851994028577522e-05,
+      "loss": 4.0139,
+      "step": 3470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.851780763489016e-05,
+      "loss": 4.0982,
+      "step": 3475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.851567498400512e-05,
+      "loss": 4.1625,
+      "step": 3480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.851354233312007e-05,
+      "loss": 4.0794,
+      "step": 3485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.851140968223503e-05,
+      "loss": 3.9972,
+      "step": 3490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.850927703134998e-05,
+      "loss": 4.1615,
+      "step": 3495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.850714438046492e-05,
+      "loss": 4.099,
+      "step": 3500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.850501172957988e-05,
+      "loss": 4.2062,
+      "step": 3505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.850287907869482e-05,
+      "loss": 4.0242,
+      "step": 3510
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.850074642780977e-05,
+      "loss": 4.114,
+      "step": 3515
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849861377692472e-05,
+      "loss": 4.1395,
+      "step": 3520
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849648112603967e-05,
+      "loss": 4.0926,
+      "step": 3525
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849434847515461e-05,
+      "loss": 4.0642,
+      "step": 3530
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849221582426958e-05,
+      "loss": 4.1643,
+      "step": 3535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849008317338452e-05,
+      "loss": 4.0857,
+      "step": 3540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.848795052249948e-05,
+      "loss": 4.0544,
+      "step": 3545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.848581787161442e-05,
+      "loss": 4.0275,
+      "step": 3550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.848368522072937e-05,
+      "loss": 4.0562,
+      "step": 3555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.848155256984433e-05,
+      "loss": 4.0621,
+      "step": 3560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.847941991895927e-05,
+      "loss": 4.1615,
+      "step": 3565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.847728726807422e-05,
+      "loss": 4.0546,
+      "step": 3570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.847515461718916e-05,
+      "loss": 4.0172,
+      "step": 3575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.847302196630412e-05,
+      "loss": 4.1494,
+      "step": 3580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.847088931541907e-05,
+      "loss": 4.0837,
+      "step": 3585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.846875666453403e-05,
+      "loss": 4.0739,
+      "step": 3590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.846662401364897e-05,
+      "loss": 4.1764,
+      "step": 3595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.846449136276392e-05,
+      "loss": 4.005,
+      "step": 3600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.846235871187887e-05,
+      "loss": 4.1376,
+      "step": 3605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.846022606099382e-05,
+      "loss": 4.05,
+      "step": 3610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.845809341010876e-05,
+      "loss": 4.0567,
+      "step": 3615
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.845596075922372e-05,
+      "loss": 4.0898,
+      "step": 3620
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.845382810833867e-05,
+      "loss": 4.0379,
+      "step": 3625
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.845169545745363e-05,
+      "loss": 4.0889,
+      "step": 3630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.844956280656858e-05,
+      "loss": 3.9299,
+      "step": 3635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.844743015568352e-05,
+      "loss": 4.0745,
+      "step": 3640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.844529750479848e-05,
+      "loss": 4.0128,
+      "step": 3645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.844316485391342e-05,
+      "loss": 4.0397,
+      "step": 3650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.844103220302837e-05,
+      "loss": 3.9828,
+      "step": 3655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.843889955214331e-05,
+      "loss": 4.0789,
+      "step": 3660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.843676690125827e-05,
+      "loss": 3.9853,
+      "step": 3665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.843463425037321e-05,
+      "loss": 4.0654,
+      "step": 3670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.843250159948816e-05,
+      "loss": 4.1018,
+      "step": 3675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.843036894860312e-05,
+      "loss": 4.1252,
+      "step": 3680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.842823629771807e-05,
+      "loss": 3.99,
+      "step": 3685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.842610364683303e-05,
+      "loss": 4.0717,
+      "step": 3690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.842397099594797e-05,
+      "loss": 4.015,
+      "step": 3695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.842183834506292e-05,
+      "loss": 4.1548,
+      "step": 3700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.841970569417787e-05,
+      "loss": 4.0925,
+      "step": 3705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.841757304329282e-05,
+      "loss": 4.0652,
+      "step": 3710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.841544039240776e-05,
+      "loss": 4.01,
+      "step": 3715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.841330774152272e-05,
+      "loss": 4.1887,
+      "step": 3720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.841117509063766e-05,
+      "loss": 3.9903,
+      "step": 3725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840904243975263e-05,
+      "loss": 4.0842,
+      "step": 3730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840690978886757e-05,
+      "loss": 4.1217,
+      "step": 3735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840477713798252e-05,
+      "loss": 4.1292,
+      "step": 3740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840264448709746e-05,
+      "loss": 4.0607,
+      "step": 3745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840051183621242e-05,
+      "loss": 4.0346,
+      "step": 3750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.839837918532737e-05,
+      "loss": 4.018,
+      "step": 3755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.839624653444231e-05,
+      "loss": 4.0427,
+      "step": 3760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.839411388355727e-05,
+      "loss": 3.9912,
+      "step": 3765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.839198123267221e-05,
+      "loss": 4.0487,
+      "step": 3770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.838984858178716e-05,
+      "loss": 4.0825,
+      "step": 3775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838771593090212e-05,
+      "loss": 4.0951,
+      "step": 3780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838558328001707e-05,
+      "loss": 3.9822,
+      "step": 3785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838345062913201e-05,
+      "loss": 4.0599,
+      "step": 3790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838131797824697e-05,
+      "loss": 4.1555,
+      "step": 3795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837918532736191e-05,
+      "loss": 4.0708,
+      "step": 3800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837705267647687e-05,
+      "loss": 4.1252,
+      "step": 3805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.83749200255918e-05,
+      "loss": 4.1803,
+      "step": 3810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837278737470676e-05,
+      "loss": 4.0704,
+      "step": 3815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837065472382172e-05,
+      "loss": 4.0226,
+      "step": 3820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.836852207293667e-05,
+      "loss": 4.0557,
+      "step": 3825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.836638942205163e-05,
+      "loss": 4.1036,
+      "step": 3830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.836425677116657e-05,
+      "loss": 4.1469,
+      "step": 3835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.836212412028152e-05,
+      "loss": 4.0344,
+      "step": 3840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835999146939646e-05,
+      "loss": 4.101,
+      "step": 3845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835785881851142e-05,
+      "loss": 4.1505,
+      "step": 3850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835572616762636e-05,
+      "loss": 4.1152,
+      "step": 3855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835359351674131e-05,
+      "loss": 4.004,
+      "step": 3860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835146086585625e-05,
+      "loss": 4.0193,
+      "step": 3865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834932821497121e-05,
+      "loss": 4.0349,
+      "step": 3870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834719556408616e-05,
+      "loss": 4.078,
+      "step": 3875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834506291320112e-05,
+      "loss": 4.0406,
+      "step": 3880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834293026231607e-05,
+      "loss": 4.0963,
+      "step": 3885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834079761143101e-05,
+      "loss": 4.0159,
+      "step": 3890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.833866496054597e-05,
+      "loss": 4.1063,
+      "step": 3895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.833653230966091e-05,
+      "loss": 4.0392,
+      "step": 3900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.833439965877587e-05,
+      "loss": 4.1441,
+      "step": 3905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.83322670078908e-05,
+      "loss": 4.1397,
+      "step": 3910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.833013435700576e-05,
+      "loss": 4.0371,
+      "step": 3915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.83280017061207e-05,
+      "loss": 4.0455,
+      "step": 3920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.832586905523567e-05,
+      "loss": 3.9925,
+      "step": 3925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.832373640435061e-05,
+      "loss": 4.1464,
+      "step": 3930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.832160375346557e-05,
+      "loss": 4.0579,
+      "step": 3935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831947110258051e-05,
+      "loss": 4.1011,
+      "step": 3940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831733845169546e-05,
+      "loss": 3.9529,
+      "step": 3945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831520580081042e-05,
+      "loss": 4.0772,
+      "step": 3950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831307314992536e-05,
+      "loss": 4.0175,
+      "step": 3955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831094049904031e-05,
+      "loss": 4.1711,
+      "step": 3960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.830880784815525e-05,
+      "loss": 4.074,
+      "step": 3965
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.830667519727021e-05,
+      "loss": 4.0718,
+      "step": 3970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.830454254638516e-05,
+      "loss": 4.0085,
+      "step": 3975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.830240989550012e-05,
+      "loss": 4.0626,
+      "step": 3980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.830027724461506e-05,
+      "loss": 3.887,
+      "step": 3985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.829814459373001e-05,
+      "loss": 4.0241,
+      "step": 3990
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.829601194284496e-05,
+      "loss": 4.0986,
+      "step": 3995
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.829387929195991e-05,
+      "loss": 4.0039,
+      "step": 4000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.829174664107485e-05,
+      "loss": 4.0547,
+      "step": 4005
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.82896139901898e-05,
+      "loss": 4.1122,
+      "step": 4010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828748133930476e-05,
+      "loss": 4.0487,
+      "step": 4015
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828534868841972e-05,
+      "loss": 3.945,
+      "step": 4020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828321603753467e-05,
+      "loss": 4.1208,
+      "step": 4025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828108338664961e-05,
+      "loss": 4.1426,
+      "step": 4030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827895073576457e-05,
+      "loss": 3.9897,
+      "step": 4035
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827681808487951e-05,
+      "loss": 4.0565,
+      "step": 4040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827468543399446e-05,
+      "loss": 4.0652,
+      "step": 4045
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.82725527831094e-05,
+      "loss": 4.1183,
+      "step": 4050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827042013222436e-05,
+      "loss": 3.9832,
+      "step": 4055
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.82682874813393e-05,
+      "loss": 4.0783,
+      "step": 4060
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.826615483045425e-05,
+      "loss": 4.0811,
+      "step": 4065
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.826402217956921e-05,
+      "loss": 4.042,
+      "step": 4070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.826188952868416e-05,
+      "loss": 4.1294,
+      "step": 4075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.825975687779912e-05,
+      "loss": 4.0604,
+      "step": 4080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.825762422691406e-05,
+      "loss": 4.1075,
+      "step": 4085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.825549157602901e-05,
+      "loss": 4.07,
+      "step": 4090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.825335892514396e-05,
+      "loss": 4.0523,
+      "step": 4095
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.825122627425891e-05,
+      "loss": 4.0995,
+      "step": 4100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.824909362337385e-05,
+      "loss": 3.9715,
+      "step": 4105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.82469609724888e-05,
+      "loss": 3.9682,
+      "step": 4110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.824482832160375e-05,
+      "loss": 4.0932,
+      "step": 4115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.824269567071872e-05,
+      "loss": 4.0342,
+      "step": 4120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.824056301983366e-05,
+      "loss": 4.0012,
+      "step": 4125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.823843036894861e-05,
+      "loss": 4.1553,
+      "step": 4130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.823629771806355e-05,
+      "loss": 4.0861,
+      "step": 4135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.823416506717851e-05,
+      "loss": 4.0703,
+      "step": 4140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.823203241629346e-05,
+      "loss": 4.0106,
+      "step": 4145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.82298997654084e-05,
+      "loss": 4.0657,
+      "step": 4150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.822776711452336e-05,
+      "loss": 4.0908,
+      "step": 4155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.82256344636383e-05,
+      "loss": 4.0279,
+      "step": 4160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.822350181275325e-05,
+      "loss": 4.027,
+      "step": 4165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.822136916186821e-05,
+      "loss": 3.9981,
+      "step": 4170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.821923651098316e-05,
+      "loss": 4.0162,
+      "step": 4175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.82171038600981e-05,
+      "loss": 4.078,
+      "step": 4180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.821497120921306e-05,
+      "loss": 4.1236,
+      "step": 4185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.8212838558328e-05,
+      "loss": 3.9386,
+      "step": 4190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.821070590744296e-05,
+      "loss": 3.9884,
+      "step": 4195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.82085732565579e-05,
+      "loss": 4.0046,
+      "step": 4200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.820644060567285e-05,
+      "loss": 3.9942,
+      "step": 4205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.82043079547878e-05,
+      "loss": 3.9106,
+      "step": 4210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.820217530390276e-05,
+      "loss": 4.0357,
+      "step": 4215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.820004265301772e-05,
+      "loss": 4.0835,
+      "step": 4220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.819791000213266e-05,
+      "loss": 4.0181,
+      "step": 4225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.819577735124761e-05,
+      "loss": 4.059,
+      "step": 4230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.819364470036255e-05,
+      "loss": 3.9714,
+      "step": 4235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.819151204947751e-05,
+      "loss": 3.9571,
+      "step": 4240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.818937939859245e-05,
+      "loss": 4.0431,
+      "step": 4245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.81872467477074e-05,
+      "loss": 4.0491,
+      "step": 4250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.818511409682235e-05,
+      "loss": 4.0191,
+      "step": 4255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.81829814459373e-05,
+      "loss": 4.0263,
+      "step": 4260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.818084879505225e-05,
+      "loss": 4.0452,
+      "step": 4265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.817871614416721e-05,
+      "loss": 3.9861,
+      "step": 4270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.817658349328216e-05,
+      "loss": 4.0039,
+      "step": 4275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.81744508423971e-05,
+      "loss": 3.9324,
+      "step": 4280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.817231819151206e-05,
+      "loss": 4.0597,
+      "step": 4285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.8170185540627e-05,
+      "loss": 4.1195,
+      "step": 4290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.816805288974196e-05,
+      "loss": 4.0344,
+      "step": 4295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.81659202388569e-05,
+      "loss": 4.0057,
+      "step": 4300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.816378758797185e-05,
+      "loss": 4.0535,
+      "step": 4305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.81616549370868e-05,
+      "loss": 4.1324,
+      "step": 4310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.815952228620176e-05,
+      "loss": 4.0478,
+      "step": 4315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.81573896353167e-05,
+      "loss": 4.1411,
+      "step": 4320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.815525698443166e-05,
+      "loss": 4.0773,
+      "step": 4325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.81531243335466e-05,
+      "loss": 4.0551,
+      "step": 4330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.815099168266155e-05,
+      "loss": 4.1022,
+      "step": 4335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.814885903177651e-05,
+      "loss": 4.0866,
+      "step": 4340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.814672638089145e-05,
+      "loss": 3.9278,
+      "step": 4345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.81445937300064e-05,
+      "loss": 4.1098,
+      "step": 4350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.814246107912135e-05,
+      "loss": 3.9803,
+      "step": 4355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.81403284282363e-05,
+      "loss": 3.9907,
+      "step": 4360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.813819577735125e-05,
+      "loss": 3.996,
+      "step": 4365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.813606312646621e-05,
+      "loss": 4.108,
+      "step": 4370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.813393047558115e-05,
+      "loss": 4.0114,
+      "step": 4375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.81317978246961e-05,
+      "loss": 4.0698,
+      "step": 4380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.812966517381105e-05,
+      "loss": 4.0204,
+      "step": 4385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.8127532522926e-05,
+      "loss": 4.11,
+      "step": 4390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.812539987204094e-05,
+      "loss": 4.1124,
+      "step": 4395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.81232672211559e-05,
+      "loss": 4.023,
+      "step": 4400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.812113457027085e-05,
+      "loss": 4.0111,
+      "step": 4405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.81190019193858e-05,
+      "loss": 4.0778,
+      "step": 4410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.811686926850076e-05,
+      "loss": 3.9801,
+      "step": 4415
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.81147366176157e-05,
+      "loss": 4.0222,
+      "step": 4420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.811260396673066e-05,
+      "loss": 4.0027,
+      "step": 4425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.81104713158456e-05,
+      "loss": 4.0564,
+      "step": 4430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.810833866496055e-05,
+      "loss": 4.0356,
+      "step": 4435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.81062060140755e-05,
+      "loss": 4.0885,
+      "step": 4440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.810407336319045e-05,
+      "loss": 4.0675,
+      "step": 4445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.810194071230539e-05,
+      "loss": 4.1067,
+      "step": 4450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.809980806142035e-05,
+      "loss": 4.0227,
+      "step": 4455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.80976754105353e-05,
+      "loss": 4.1558,
+      "step": 4460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.809554275965025e-05,
+      "loss": 4.0434,
+      "step": 4465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.809341010876521e-05,
+      "loss": 3.9757,
+      "step": 4470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.809127745788015e-05,
+      "loss": 4.0859,
+      "step": 4475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.80891448069951e-05,
+      "loss": 4.1458,
+      "step": 4480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808701215611005e-05,
+      "loss": 3.9807,
+      "step": 4485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.8084879505225e-05,
+      "loss": 4.1011,
+      "step": 4490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808274685433994e-05,
+      "loss": 4.1034,
+      "step": 4495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.80806142034549e-05,
+      "loss": 4.0167,
+      "step": 4500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.807848155256985e-05,
+      "loss": 4.0626,
+      "step": 4505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.80763489016848e-05,
+      "loss": 3.9839,
+      "step": 4510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.807421625079975e-05,
+      "loss": 4.0531,
+      "step": 4515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.80720835999147e-05,
+      "loss": 4.0065,
+      "step": 4520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.806995094902964e-05,
+      "loss": 4.0236,
+      "step": 4525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.80678182981446e-05,
+      "loss": 3.9416,
+      "step": 4530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.806568564725955e-05,
+      "loss": 4.0856,
+      "step": 4535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.80635529963745e-05,
+      "loss": 4.0172,
+      "step": 4540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.806142034548945e-05,
+      "loss": 4.0182,
+      "step": 4545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.805928769460439e-05,
+      "loss": 4.0691,
+      "step": 4550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.805715504371935e-05,
+      "loss": 3.9655,
+      "step": 4555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.80550223928343e-05,
+      "loss": 4.088,
+      "step": 4560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.805288974194925e-05,
+      "loss": 4.1289,
+      "step": 4565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.80507570910642e-05,
+      "loss": 4.0653,
+      "step": 4570
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.804862444017915e-05,
+      "loss": 4.0539,
+      "step": 4575
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.804649178929409e-05,
+      "loss": 4.0928,
+      "step": 4580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.804435913840905e-05,
+      "loss": 3.9413,
+      "step": 4585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.804222648752399e-05,
+      "loss": 4.1666,
+      "step": 4590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.804009383663894e-05,
+      "loss": 4.1192,
+      "step": 4595
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.80379611857539e-05,
+      "loss": 3.9901,
+      "step": 4600
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.803582853486885e-05,
+      "loss": 4.0651,
+      "step": 4605
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.80336958839838e-05,
+      "loss": 3.9998,
+      "step": 4610
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.803156323309875e-05,
+      "loss": 4.1184,
+      "step": 4615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.80294305822137e-05,
+      "loss": 4.0873,
+      "step": 4620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.802729793132864e-05,
+      "loss": 4.0269,
+      "step": 4625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.80251652804436e-05,
+      "loss": 4.0547,
+      "step": 4630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.802303262955854e-05,
+      "loss": 4.0636,
+      "step": 4635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.80208999786735e-05,
+      "loss": 4.0391,
+      "step": 4640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.801876732778844e-05,
+      "loss": 4.0591,
+      "step": 4645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.801663467690339e-05,
+      "loss": 4.0858,
+      "step": 4650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.801450202601835e-05,
+      "loss": 3.9863,
+      "step": 4655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.80123693751333e-05,
+      "loss": 4.0796,
+      "step": 4660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.801023672424825e-05,
+      "loss": 4.0941,
+      "step": 4665
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.80081040733632e-05,
+      "loss": 3.9785,
+      "step": 4670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.800597142247815e-05,
+      "loss": 4.0741,
+      "step": 4675
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.800383877159309e-05,
+      "loss": 4.0798,
+      "step": 4680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.800170612070805e-05,
+      "loss": 4.0571,
+      "step": 4685
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.799957346982299e-05,
+      "loss": 4.1389,
+      "step": 4690
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.799744081893794e-05,
+      "loss": 4.0451,
+      "step": 4695
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.79953081680529e-05,
+      "loss": 4.0766,
+      "step": 4700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.799317551716785e-05,
+      "loss": 4.0352,
+      "step": 4705
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.799104286628279e-05,
+      "loss": 4.034,
+      "step": 4710
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.798891021539775e-05,
+      "loss": 4.1231,
+      "step": 4715
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.798677756451269e-05,
+      "loss": 4.0251,
+      "step": 4720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.798464491362764e-05,
+      "loss": 4.0337,
+      "step": 4725
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.79825122627426e-05,
+      "loss": 3.9813,
+      "step": 4730
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.798037961185754e-05,
+      "loss": 4.0151,
+      "step": 4735
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.79782469609725e-05,
+      "loss": 4.0088,
+      "step": 4740
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.797611431008744e-05,
+      "loss": 4.0241,
+      "step": 4745
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.797398165920239e-05,
+      "loss": 3.9885,
+      "step": 4750
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.797184900831735e-05,
+      "loss": 4.0564,
+      "step": 4755
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.79697163574323e-05,
+      "loss": 4.1148,
+      "step": 4760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.796758370654724e-05,
+      "loss": 3.9412,
+      "step": 4765
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.79654510556622e-05,
+      "loss": 3.9033,
+      "step": 4770
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.796331840477714e-05,
+      "loss": 4.1227,
+      "step": 4775
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.796118575389209e-05,
+      "loss": 4.0332,
+      "step": 4780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.795905310300705e-05,
+      "loss": 4.0095,
+      "step": 4785
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.795692045212199e-05,
+      "loss": 4.0263,
+      "step": 4790
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.795478780123694e-05,
+      "loss": 4.0743,
+      "step": 4795
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.79526551503519e-05,
+      "loss": 4.0627,
+      "step": 4800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.795052249946685e-05,
+      "loss": 4.0533,
+      "step": 4805
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.79483898485818e-05,
+      "loss": 4.0252,
+      "step": 4810
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.794625719769675e-05,
+      "loss": 3.9966,
+      "step": 4815
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.794412454681169e-05,
+      "loss": 3.9786,
+      "step": 4820
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.794199189592664e-05,
+      "loss": 4.0399,
+      "step": 4825
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.793985924504159e-05,
+      "loss": 3.9489,
+      "step": 4830
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.793772659415654e-05,
+      "loss": 4.1798,
+      "step": 4835
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.793559394327148e-05,
+      "loss": 4.0606,
+      "step": 4840
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.793346129238644e-05,
+      "loss": 4.0424,
+      "step": 4845
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.793132864150139e-05,
+      "loss": 3.984,
+      "step": 4850
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.792919599061635e-05,
+      "loss": 4.0398,
+      "step": 4855
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.79270633397313e-05,
+      "loss": 3.9847,
+      "step": 4860
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.792493068884624e-05,
+      "loss": 4.0735,
+      "step": 4865
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.79227980379612e-05,
+      "loss": 4.0534,
+      "step": 4870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.792066538707614e-05,
+      "loss": 4.0098,
+      "step": 4875
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791853273619109e-05,
+      "loss": 4.0024,
+      "step": 4880
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791640008530603e-05,
+      "loss": 4.0428,
+      "step": 4885
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791426743442099e-05,
+      "loss": 3.9316,
+      "step": 4890
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791213478353594e-05,
+      "loss": 4.0487,
+      "step": 4895
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.79100021326509e-05,
+      "loss": 3.9686,
+      "step": 4900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.790786948176584e-05,
+      "loss": 4.0,
+      "step": 4905
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.79057368308808e-05,
+      "loss": 4.0191,
+      "step": 4910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.790360417999575e-05,
+      "loss": 3.8967,
+      "step": 4915
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.790147152911069e-05,
+      "loss": 4.0222,
+      "step": 4920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.789933887822564e-05,
+      "loss": 3.9386,
+      "step": 4925
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.789720622734059e-05,
+      "loss": 4.0404,
+      "step": 4930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.789507357645554e-05,
+      "loss": 4.066,
+      "step": 4935
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.789294092557048e-05,
+      "loss": 3.9719,
+      "step": 4940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.789080827468544e-05,
+      "loss": 4.0497,
+      "step": 4945
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.788867562380039e-05,
+      "loss": 4.0181,
+      "step": 4950
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.788654297291535e-05,
+      "loss": 4.0261,
+      "step": 4955
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.788441032203029e-05,
+      "loss": 4.0205,
+      "step": 4960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.788227767114524e-05,
+      "loss": 3.9148,
+      "step": 4965
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.788014502026018e-05,
+      "loss": 3.9803,
+      "step": 4970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.787801236937514e-05,
+      "loss": 3.9548,
+      "step": 4975
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.787587971849009e-05,
+      "loss": 4.0615,
+      "step": 4980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.787374706760503e-05,
+      "loss": 3.9432,
+      "step": 4985
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.787161441671999e-05,
+      "loss": 4.1357,
+      "step": 4990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.786948176583494e-05,
+      "loss": 3.9741,
+      "step": 4995
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.78673491149499e-05,
+      "loss": 3.9806,
+      "step": 5000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.786521646406484e-05,
+      "loss": 3.9904,
+      "step": 5005
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.78630838131798e-05,
+      "loss": 3.9437,
+      "step": 5010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.786095116229473e-05,
+      "loss": 4.0055,
+      "step": 5015
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.785881851140969e-05,
+      "loss": 4.0632,
+      "step": 5020
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.785668586052463e-05,
+      "loss": 4.0871,
+      "step": 5025
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.785455320963959e-05,
+      "loss": 4.0062,
+      "step": 5030
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.785242055875453e-05,
+      "loss": 4.1021,
+      "step": 5035
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.785028790786948e-05,
+      "loss": 4.0405,
+      "step": 5040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.784815525698444e-05,
+      "loss": 4.0284,
+      "step": 5045
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.784602260609939e-05,
+      "loss": 3.9969,
+      "step": 5050
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.784388995521435e-05,
+      "loss": 4.054,
+      "step": 5055
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.784175730432929e-05,
+      "loss": 4.0135,
+      "step": 5060
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.783962465344424e-05,
+      "loss": 4.114,
+      "step": 5065
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.783749200255918e-05,
+      "loss": 3.9327,
+      "step": 5070
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.783535935167414e-05,
+      "loss": 4.0121,
+      "step": 5075
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.783322670078908e-05,
+      "loss": 3.981,
+      "step": 5080
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.783109404990403e-05,
+      "loss": 3.9656,
+      "step": 5085
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.782896139901899e-05,
+      "loss": 3.9937,
+      "step": 5090
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.782682874813394e-05,
+      "loss": 4.0654,
+      "step": 5095
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.782469609724888e-05,
+      "loss": 4.0631,
+      "step": 5100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.782256344636384e-05,
+      "loss": 3.985,
+      "step": 5105
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.78204307954788e-05,
+      "loss": 3.9168,
+      "step": 5110
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.781829814459373e-05,
+      "loss": 4.0774,
+      "step": 5115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.781616549370869e-05,
+      "loss": 3.9991,
+      "step": 5120
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.781403284282363e-05,
+      "loss": 3.9888,
+      "step": 5125
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.781190019193859e-05,
+      "loss": 3.9953,
+      "step": 5130
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.780976754105353e-05,
+      "loss": 3.9908,
+      "step": 5135
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.780763489016848e-05,
+      "loss": 4.0206,
+      "step": 5140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.780550223928344e-05,
+      "loss": 3.9307,
+      "step": 5145
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.780336958839839e-05,
+      "loss": 3.963,
+      "step": 5150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.780123693751333e-05,
+      "loss": 4.0101,
+      "step": 5155
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.779910428662829e-05,
+      "loss": 3.9811,
+      "step": 5160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.779697163574323e-05,
+      "loss": 4.0322,
+      "step": 5165
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.779483898485818e-05,
+      "loss": 3.9802,
+      "step": 5170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.779270633397314e-05,
+      "loss": 4.0302,
+      "step": 5175
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.779057368308808e-05,
+      "loss": 3.8904,
+      "step": 5180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.778844103220303e-05,
+      "loss": 4.1,
+      "step": 5185
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.778630838131799e-05,
+      "loss": 4.0124,
+      "step": 5190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.778417573043294e-05,
+      "loss": 3.9944,
+      "step": 5195
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.778204307954788e-05,
+      "loss": 4.0364,
+      "step": 5200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.777991042866284e-05,
+      "loss": 4.0015,
+      "step": 5205
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.777777777777778e-05,
+      "loss": 4.0459,
+      "step": 5210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.777564512689273e-05,
+      "loss": 3.9391,
+      "step": 5215
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.777351247600768e-05,
+      "loss": 3.9457,
+      "step": 5220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.777137982512263e-05,
+      "loss": 4.0463,
+      "step": 5225
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776924717423757e-05,
+      "loss": 3.9905,
+      "step": 5230
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776711452335253e-05,
+      "loss": 4.0013,
+      "step": 5235
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776498187246748e-05,
+      "loss": 3.9897,
+      "step": 5240
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776284922158244e-05,
+      "loss": 4.0325,
+      "step": 5245
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776071657069739e-05,
+      "loss": 3.965,
+      "step": 5250
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.775858391981233e-05,
+      "loss": 4.0412,
+      "step": 5255
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.775645126892729e-05,
+      "loss": 3.9602,
+      "step": 5260
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.775431861804223e-05,
+      "loss": 4.1305,
+      "step": 5265
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.775218596715718e-05,
+      "loss": 3.9748,
+      "step": 5270
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.775005331627212e-05,
+      "loss": 4.0559,
+      "step": 5275
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.774792066538708e-05,
+      "loss": 3.9368,
+      "step": 5280
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.774578801450203e-05,
+      "loss": 4.027,
+      "step": 5285
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.774365536361699e-05,
+      "loss": 3.9226,
+      "step": 5290
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.774152271273193e-05,
+      "loss": 4.0619,
+      "step": 5295
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.773939006184688e-05,
+      "loss": 3.9517,
+      "step": 5300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.773725741096184e-05,
+      "loss": 4.0764,
+      "step": 5305
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.773512476007678e-05,
+      "loss": 4.0074,
+      "step": 5310
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.773299210919173e-05,
+      "loss": 4.0077,
+      "step": 5315
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.773085945830668e-05,
+      "loss": 3.9018,
+      "step": 5320
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.772872680742163e-05,
+      "loss": 3.9531,
+      "step": 5325
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.772659415653657e-05,
+      "loss": 4.0914,
+      "step": 5330
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.772446150565153e-05,
+      "loss": 3.9734,
+      "step": 5335
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.772232885476648e-05,
+      "loss": 4.063,
+      "step": 5340
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.772019620388144e-05,
+      "loss": 4.0108,
+      "step": 5345
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.771806355299638e-05,
+      "loss": 4.1145,
+      "step": 5350
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.771593090211133e-05,
+      "loss": 3.965,
+      "step": 5355
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.771379825122627e-05,
+      "loss": 3.9633,
+      "step": 5360
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.771166560034123e-05,
+      "loss": 3.8689,
+      "step": 5365
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.770953294945618e-05,
+      "loss": 3.9957,
+      "step": 5370
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.770740029857112e-05,
+      "loss": 3.9757,
+      "step": 5375
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.770526764768608e-05,
+      "loss": 4.0402,
+      "step": 5380
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.770313499680103e-05,
+      "loss": 3.9377,
+      "step": 5385
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.770100234591599e-05,
+      "loss": 3.9933,
+      "step": 5390
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.769886969503093e-05,
+      "loss": 3.9787,
+      "step": 5395
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.769673704414588e-05,
+      "loss": 4.0208,
+      "step": 5400
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.769460439326083e-05,
+      "loss": 4.0543,
+      "step": 5405
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.769247174237578e-05,
+      "loss": 4.0265,
+      "step": 5410
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.769033909149072e-05,
+      "loss": 4.0181,
+      "step": 5415
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.768820644060568e-05,
+      "loss": 4.0528,
+      "step": 5420
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.768607378972062e-05,
+      "loss": 4.0293,
+      "step": 5425
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.768394113883557e-05,
+      "loss": 3.9797,
+      "step": 5430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.768180848795053e-05,
+      "loss": 4.0331,
+      "step": 5435
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.767967583706548e-05,
+      "loss": 3.9143,
+      "step": 5440
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.767754318618044e-05,
+      "loss": 4.0379,
+      "step": 5445
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.767541053529538e-05,
+      "loss": 3.9779,
+      "step": 5450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.767327788441033e-05,
+      "loss": 3.9843,
+      "step": 5455
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.767114523352527e-05,
+      "loss": 4.0985,
+      "step": 5460
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.766901258264023e-05,
+      "loss": 4.0878,
+      "step": 5465
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.766687993175517e-05,
+      "loss": 3.9589,
+      "step": 5470
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.766474728087012e-05,
+      "loss": 4.0069,
+      "step": 5475
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.766261462998508e-05,
+      "loss": 4.0183,
+      "step": 5480
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.766048197910003e-05,
+      "loss": 4.0204,
+      "step": 5485
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.765834932821497e-05,
+      "loss": 4.027,
+      "step": 5490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.765621667732993e-05,
+      "loss": 4.0197,
+      "step": 5495
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.765408402644488e-05,
+      "loss": 3.882,
+      "step": 5500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.765195137555983e-05,
+      "loss": 4.0029,
+      "step": 5505
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.764981872467478e-05,
+      "loss": 3.9696,
+      "step": 5510
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.764768607378972e-05,
+      "loss": 4.0449,
+      "step": 5515
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.764555342290468e-05,
+      "loss": 4.0341,
+      "step": 5520
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.764342077201962e-05,
+      "loss": 3.9208,
+      "step": 5525
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.764128812113457e-05,
+      "loss": 4.0587,
+      "step": 5530
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.763915547024953e-05,
+      "loss": 4.0061,
+      "step": 5535
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.763702281936448e-05,
+      "loss": 3.9587,
+      "step": 5540
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.763489016847942e-05,
+      "loss": 4.0801,
+      "step": 5545
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.763275751759438e-05,
+      "loss": 4.0097,
+      "step": 5550
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.763062486670932e-05,
+      "loss": 3.9811,
+      "step": 5555
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.762849221582427e-05,
+      "loss": 4.0808,
+      "step": 5560
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.762635956493923e-05,
+      "loss": 4.0921,
+      "step": 5565
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.762422691405417e-05,
+      "loss": 3.8761,
+      "step": 5570
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.762209426316912e-05,
+      "loss": 3.9694,
+      "step": 5575
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.761996161228408e-05,
+      "loss": 3.9674,
+      "step": 5580
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.761782896139903e-05,
+      "loss": 4.0061,
+      "step": 5585
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.761569631051397e-05,
+      "loss": 3.9304,
+      "step": 5590
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.761356365962893e-05,
+      "loss": 3.9911,
+      "step": 5595
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.761143100874387e-05,
+      "loss": 3.9715,
+      "step": 5600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.760929835785883e-05,
+      "loss": 3.9093,
+      "step": 5605
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.760716570697377e-05,
+      "loss": 3.9688,
+      "step": 5610
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.760503305608872e-05,
+      "loss": 3.9166,
+      "step": 5615
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.760290040520366e-05,
+      "loss": 4.1225,
+      "step": 5620
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.760076775431862e-05,
+      "loss": 4.0253,
+      "step": 5625
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.759863510343357e-05,
+      "loss": 3.9521,
+      "step": 5630
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.759650245254853e-05,
+      "loss": 4.0724,
+      "step": 5635
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.759436980166348e-05,
+      "loss": 4.0403,
+      "step": 5640
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.759223715077842e-05,
+      "loss": 4.006,
+      "step": 5645
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.759010449989338e-05,
+      "loss": 3.9746,
+      "step": 5650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.758797184900832e-05,
+      "loss": 4.0022,
+      "step": 5655
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.758583919812327e-05,
+      "loss": 4.0028,
+      "step": 5660
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.758370654723821e-05,
+      "loss": 4.0601,
+      "step": 5665
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.758157389635317e-05,
+      "loss": 3.9339,
+      "step": 5670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.757944124546812e-05,
+      "loss": 3.9979,
+      "step": 5675
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.757730859458308e-05,
+      "loss": 3.917,
+      "step": 5680
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.757517594369802e-05,
+      "loss": 3.9101,
+      "step": 5685
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.757304329281297e-05,
+      "loss": 3.9277,
+      "step": 5690
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.757091064192793e-05,
+      "loss": 3.9479,
+      "step": 5695
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.756877799104287e-05,
+      "loss": 3.9229,
+      "step": 5700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.756664534015783e-05,
+      "loss": 3.9866,
+      "step": 5705
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.756451268927277e-05,
+      "loss": 4.0105,
+      "step": 5710
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.756238003838772e-05,
+      "loss": 4.0302,
+      "step": 5715
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.756024738750266e-05,
+      "loss": 3.9889,
+      "step": 5720
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.755811473661762e-05,
+      "loss": 4.0629,
+      "step": 5725
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.755598208573257e-05,
+      "loss": 3.9892,
+      "step": 5730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.755384943484753e-05,
+      "loss": 3.9448,
+      "step": 5735
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.755171678396247e-05,
+      "loss": 3.9498,
+      "step": 5740
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.754958413307742e-05,
+      "loss": 4.0775,
+      "step": 5745
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.754745148219236e-05,
+      "loss": 3.977,
+      "step": 5750
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.754531883130732e-05,
+      "loss": 3.9594,
+      "step": 5755
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.754318618042227e-05,
+      "loss": 4.0098,
+      "step": 5760
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.754105352953721e-05,
+      "loss": 4.0366,
+      "step": 5765
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.753892087865217e-05,
+      "loss": 3.9989,
+      "step": 5770
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.753678822776712e-05,
+      "loss": 4.0615,
+      "step": 5775
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.753465557688208e-05,
+      "loss": 3.935,
+      "step": 5780
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.753252292599702e-05,
+      "loss": 4.0137,
+      "step": 5785
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.753039027511197e-05,
+      "loss": 4.015,
+      "step": 5790
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.752825762422692e-05,
+      "loss": 4.0238,
+      "step": 5795
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.752612497334187e-05,
+      "loss": 3.9634,
+      "step": 5800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.752399232245681e-05,
+      "loss": 3.8972,
+      "step": 5805
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.752185967157177e-05,
+      "loss": 3.911,
+      "step": 5810
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.751972702068671e-05,
+      "loss": 3.9551,
+      "step": 5815
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.751759436980166e-05,
+      "loss": 3.9579,
+      "step": 5820
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.751546171891662e-05,
+      "loss": 4.0427,
+      "step": 5825
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.751332906803157e-05,
+      "loss": 3.9728,
+      "step": 5830
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.751119641714653e-05,
+      "loss": 3.938,
+      "step": 5835
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750906376626147e-05,
+      "loss": 3.99,
+      "step": 5840
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750693111537642e-05,
+      "loss": 4.0078,
+      "step": 5845
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750479846449136e-05,
+      "loss": 3.9465,
+      "step": 5850
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750266581360632e-05,
+      "loss": 3.9558,
+      "step": 5855
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750053316272126e-05,
+      "loss": 4.0207,
+      "step": 5860
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.749840051183621e-05,
+      "loss": 3.9998,
+      "step": 5865
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.749626786095117e-05,
+      "loss": 3.9949,
+      "step": 5870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.749413521006612e-05,
+      "loss": 4.1192,
+      "step": 5875
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.749200255918106e-05,
+      "loss": 4.0863,
+      "step": 5880
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.748986990829602e-05,
+      "loss": 3.9634,
+      "step": 5885
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.748773725741097e-05,
+      "loss": 4.0147,
+      "step": 5890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.748560460652592e-05,
+      "loss": 3.9261,
+      "step": 5895
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.748347195564087e-05,
+      "loss": 3.9583,
+      "step": 5900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.748133930475581e-05,
+      "loss": 3.9653,
+      "step": 5905
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.747920665387077e-05,
+      "loss": 3.9179,
+      "step": 5910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.747707400298571e-05,
+      "loss": 4.0832,
+      "step": 5915
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.747494135210066e-05,
+      "loss": 3.9294,
+      "step": 5920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.747280870121562e-05,
+      "loss": 3.9206,
+      "step": 5925
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.747067605033057e-05,
+      "loss": 4.0002,
+      "step": 5930
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.746854339944551e-05,
+      "loss": 3.9828,
+      "step": 5935
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.746641074856047e-05,
+      "loss": 3.9561,
+      "step": 5940
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.746427809767541e-05,
+      "loss": 4.0734,
+      "step": 5945
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.746214544679036e-05,
+      "loss": 3.9524,
+      "step": 5950
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.746001279590532e-05,
+      "loss": 3.9586,
+      "step": 5955
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.745788014502026e-05,
+      "loss": 3.9722,
+      "step": 5960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.745574749413521e-05,
+      "loss": 3.9812,
+      "step": 5965
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.745361484325017e-05,
+      "loss": 3.9916,
+      "step": 5970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.745148219236512e-05,
+      "loss": 4.072,
+      "step": 5975
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744934954148006e-05,
+      "loss": 3.9842,
+      "step": 5980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744721689059502e-05,
+      "loss": 3.938,
+      "step": 5985
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744508423970996e-05,
+      "loss": 4.1343,
+      "step": 5990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744295158882492e-05,
+      "loss": 3.9017,
+      "step": 5995
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744081893793986e-05,
+      "loss": 3.8578,
+      "step": 6000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.743868628705481e-05,
+      "loss": 3.94,
+      "step": 6005
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.743655363616975e-05,
+      "loss": 3.9782,
+      "step": 6010
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.743442098528471e-05,
+      "loss": 3.9211,
+      "step": 6015
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.743228833439966e-05,
+      "loss": 3.9576,
+      "step": 6020
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.743015568351462e-05,
+      "loss": 4.0611,
+      "step": 6025
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.742802303262957e-05,
+      "loss": 4.0018,
+      "step": 6030
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.742589038174451e-05,
+      "loss": 3.8432,
+      "step": 6035
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.742375773085947e-05,
+      "loss": 3.9131,
+      "step": 6040
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.742162507997441e-05,
+      "loss": 4.0715,
+      "step": 6045
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.741949242908936e-05,
+      "loss": 3.9914,
+      "step": 6050
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.74173597782043e-05,
+      "loss": 3.8975,
+      "step": 6055
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.741522712731926e-05,
+      "loss": 3.9765,
+      "step": 6060
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.741309447643421e-05,
+      "loss": 4.0488,
+      "step": 6065
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.741096182554917e-05,
+      "loss": 3.9656,
+      "step": 6070
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.740882917466411e-05,
+      "loss": 4.0355,
+      "step": 6075
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.740669652377907e-05,
+      "loss": 4.0277,
+      "step": 6080
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.740456387289402e-05,
+      "loss": 3.9381,
+      "step": 6085
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.740243122200896e-05,
+      "loss": 3.9785,
+      "step": 6090
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.740029857112392e-05,
+      "loss": 3.8608,
+      "step": 6095
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.739816592023886e-05,
+      "loss": 4.0212,
+      "step": 6100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.739603326935381e-05,
+      "loss": 3.8705,
+      "step": 6105
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.739390061846875e-05,
+      "loss": 4.0517,
+      "step": 6110
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.739176796758371e-05,
+      "loss": 3.9316,
+      "step": 6115
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.738963531669866e-05,
+      "loss": 3.8944,
+      "step": 6120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.738750266581362e-05,
+      "loss": 3.9119,
+      "step": 6125
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.738537001492856e-05,
+      "loss": 3.942,
+      "step": 6130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.738323736404351e-05,
+      "loss": 4.0059,
+      "step": 6135
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.738110471315845e-05,
+      "loss": 4.0799,
+      "step": 6140
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.737897206227341e-05,
+      "loss": 3.9049,
+      "step": 6145
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.737683941138836e-05,
+      "loss": 3.9992,
+      "step": 6150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.73747067605033e-05,
+      "loss": 3.9884,
+      "step": 6155
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.737257410961826e-05,
+      "loss": 4.0387,
+      "step": 6160
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.737044145873321e-05,
+      "loss": 3.9519,
+      "step": 6165
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.736830880784817e-05,
+      "loss": 4.0121,
+      "step": 6170
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.736617615696311e-05,
+      "loss": 4.0136,
+      "step": 6175
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.736404350607807e-05,
+      "loss": 3.9673,
+      "step": 6180
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.7361910855193e-05,
+      "loss": 3.971,
+      "step": 6185
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.735977820430796e-05,
+      "loss": 4.0319,
+      "step": 6190
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.73576455534229e-05,
+      "loss": 3.9679,
+      "step": 6195
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.735551290253786e-05,
+      "loss": 3.9192,
+      "step": 6200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.73533802516528e-05,
+      "loss": 3.9561,
+      "step": 6205
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.735124760076775e-05,
+      "loss": 3.9399,
+      "step": 6210
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.734911494988271e-05,
+      "loss": 3.9983,
+      "step": 6215
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.734698229899766e-05,
+      "loss": 4.0306,
+      "step": 6220
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.734484964811262e-05,
+      "loss": 3.9713,
+      "step": 6225
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.734271699722756e-05,
+      "loss": 4.0409,
+      "step": 6230
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.734058434634251e-05,
+      "loss": 4.0486,
+      "step": 6235
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.733845169545745e-05,
+      "loss": 3.9206,
+      "step": 6240
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.733631904457241e-05,
+      "loss": 3.9002,
+      "step": 6245
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.733418639368735e-05,
+      "loss": 4.0133,
+      "step": 6250
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.73320537428023e-05,
+      "loss": 4.0602,
+      "step": 6255
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.732992109191726e-05,
+      "loss": 3.931,
+      "step": 6260
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.732778844103221e-05,
+      "loss": 4.0271,
+      "step": 6265
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.732565579014716e-05,
+      "loss": 3.8892,
+      "step": 6270
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.732352313926211e-05,
+      "loss": 3.996,
+      "step": 6275
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.732139048837707e-05,
+      "loss": 3.9151,
+      "step": 6280
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.7319257837492e-05,
+      "loss": 3.9578,
+      "step": 6285
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.731712518660696e-05,
+      "loss": 4.057,
+      "step": 6290
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.73149925357219e-05,
+      "loss": 3.991,
+      "step": 6295
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.731285988483686e-05,
+      "loss": 4.0351,
+      "step": 6300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.73107272339518e-05,
+      "loss": 3.9629,
+      "step": 6305
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.730859458306675e-05,
+      "loss": 4.0369,
+      "step": 6310
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.730646193218171e-05,
+      "loss": 4.0343,
+      "step": 6315
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.730432928129666e-05,
+      "loss": 3.992,
+      "step": 6320
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.73021966304116e-05,
+      "loss": 4.0402,
+      "step": 6325
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.730006397952656e-05,
+      "loss": 3.8911,
+      "step": 6330
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.72979313286415e-05,
+      "loss": 3.9832,
+      "step": 6335
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.729579867775645e-05,
+      "loss": 4.1012,
+      "step": 6340
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.729366602687141e-05,
+      "loss": 4.0473,
+      "step": 6345
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.729153337598635e-05,
+      "loss": 3.9977,
+      "step": 6350
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.72894007251013e-05,
+      "loss": 3.9992,
+      "step": 6355
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.728726807421626e-05,
+      "loss": 3.8766,
+      "step": 6360
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.728513542333121e-05,
+      "loss": 3.9757,
+      "step": 6365
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.728300277244616e-05,
+      "loss": 4.0369,
+      "step": 6370
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.728087012156111e-05,
+      "loss": 4.0345,
+      "step": 6375
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.727873747067605e-05,
+      "loss": 4.044,
+      "step": 6380
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.7276604819791e-05,
+      "loss": 3.975,
+      "step": 6385
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.727447216890595e-05,
+      "loss": 3.9199,
+      "step": 6390
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.72723395180209e-05,
+      "loss": 4.0104,
+      "step": 6395
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.727020686713584e-05,
+      "loss": 3.996,
+      "step": 6400
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.72680742162508e-05,
+      "loss": 3.9792,
+      "step": 6405
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.726594156536575e-05,
+      "loss": 3.892,
+      "step": 6410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.726380891448071e-05,
+      "loss": 4.025,
+      "step": 6415
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.726167626359566e-05,
+      "loss": 3.9211,
+      "step": 6420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.72595436127106e-05,
+      "loss": 4.0579,
+      "step": 6425
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.725741096182556e-05,
+      "loss": 3.9146,
+      "step": 6430
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.72552783109405e-05,
+      "loss": 3.9175,
+      "step": 6435
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.725314566005545e-05,
+      "loss": 3.926,
+      "step": 6440
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.72510130091704e-05,
+      "loss": 3.9523,
+      "step": 6445
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.724888035828535e-05,
+      "loss": 3.8757,
+      "step": 6450
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.72467477074003e-05,
+      "loss": 4.0347,
+      "step": 6455
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.724461505651526e-05,
+      "loss": 3.9398,
+      "step": 6460
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.72424824056302e-05,
+      "loss": 3.9983,
+      "step": 6465
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.724034975474516e-05,
+      "loss": 3.9665,
+      "step": 6470
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.723821710386011e-05,
+      "loss": 3.8756,
+      "step": 6475
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.723608445297505e-05,
+      "loss": 4.0386,
+      "step": 6480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.723395180209e-05,
+      "loss": 4.0804,
+      "step": 6485
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.723181915120495e-05,
+      "loss": 3.967,
+      "step": 6490
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.72296865003199e-05,
+      "loss": 3.9696,
+      "step": 6495
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.722755384943484e-05,
+      "loss": 4.0244,
+      "step": 6500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.72254211985498e-05,
+      "loss": 3.9741,
+      "step": 6505
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.722328854766475e-05,
+      "loss": 3.964,
+      "step": 6510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.722115589677971e-05,
+      "loss": 4.0751,
+      "step": 6515
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.721902324589465e-05,
+      "loss": 3.9303,
+      "step": 6520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.72168905950096e-05,
+      "loss": 4.0814,
+      "step": 6525
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.721475794412454e-05,
+      "loss": 3.9309,
+      "step": 6530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.72126252932395e-05,
+      "loss": 4.0157,
+      "step": 6535
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.721049264235445e-05,
+      "loss": 3.9339,
+      "step": 6540
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.72083599914694e-05,
+      "loss": 4.0612,
+      "step": 6545
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.720622734058435e-05,
+      "loss": 4.0784,
+      "step": 6550
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.72040946896993e-05,
+      "loss": 3.9563,
+      "step": 6555
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.720196203881426e-05,
+      "loss": 3.9543,
+      "step": 6560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.71998293879292e-05,
+      "loss": 3.9408,
+      "step": 6565
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.719769673704416e-05,
+      "loss": 3.9911,
+      "step": 6570
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.71955640861591e-05,
+      "loss": 4.0082,
+      "step": 6575
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.719343143527405e-05,
+      "loss": 3.9502,
+      "step": 6580
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.719129878438899e-05,
+      "loss": 3.8762,
+      "step": 6585
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.718916613350395e-05,
+      "loss": 3.9223,
+      "step": 6590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.718703348261889e-05,
+      "loss": 3.9421,
+      "step": 6595
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.718490083173384e-05,
+      "loss": 3.9378,
+      "step": 6600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.71827681808488e-05,
+      "loss": 3.9996,
+      "step": 6605
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.718063552996375e-05,
+      "loss": 3.9914,
+      "step": 6610
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.717850287907871e-05,
+      "loss": 3.9317,
+      "step": 6615
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.717637022819365e-05,
+      "loss": 3.9648,
+      "step": 6620
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.71742375773086e-05,
+      "loss": 3.9988,
+      "step": 6625
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.717210492642354e-05,
+      "loss": 3.9859,
+      "step": 6630
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.71699722755385e-05,
+      "loss": 3.9104,
+      "step": 6635
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.716783962465344e-05,
+      "loss": 3.9238,
+      "step": 6640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.71657069737684e-05,
+      "loss": 3.9543,
+      "step": 6645
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.716357432288335e-05,
+      "loss": 3.9124,
+      "step": 6650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.71614416719983e-05,
+      "loss": 3.8599,
+      "step": 6655
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.715930902111326e-05,
+      "loss": 3.9637,
+      "step": 6660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.71571763702282e-05,
+      "loss": 3.935,
+      "step": 6665
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.715504371934316e-05,
+      "loss": 3.9602,
+      "step": 6670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.71529110684581e-05,
+      "loss": 3.9441,
+      "step": 6675
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.715077841757305e-05,
+      "loss": 4.0186,
+      "step": 6680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.714864576668799e-05,
+      "loss": 3.8798,
+      "step": 6685
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.714651311580295e-05,
+      "loss": 3.9574,
+      "step": 6690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.714438046491789e-05,
+      "loss": 3.9688,
+      "step": 6695
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.714224781403284e-05,
+      "loss": 3.9298,
+      "step": 6700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.71401151631478e-05,
+      "loss": 3.8945,
+      "step": 6705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.713798251226275e-05,
+      "loss": 3.9595,
+      "step": 6710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.71358498613777e-05,
+      "loss": 4.0051,
+      "step": 6715
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.713371721049265e-05,
+      "loss": 3.9391,
+      "step": 6720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.71315845596076e-05,
+      "loss": 3.9339,
+      "step": 6725
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.712945190872254e-05,
+      "loss": 3.9424,
+      "step": 6730
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.71273192578375e-05,
+      "loss": 4.0085,
+      "step": 6735
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.712518660695244e-05,
+      "loss": 3.9636,
+      "step": 6740
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.71230539560674e-05,
+      "loss": 3.9177,
+      "step": 6745
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.712092130518235e-05,
+      "loss": 3.9685,
+      "step": 6750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.71187886542973e-05,
+      "loss": 4.0065,
+      "step": 6755
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.711665600341225e-05,
+      "loss": 3.9373,
+      "step": 6760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.71145233525272e-05,
+      "loss": 3.969,
+      "step": 6765
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.711239070164214e-05,
+      "loss": 3.9644,
+      "step": 6770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.71102580507571e-05,
+      "loss": 3.8681,
+      "step": 6775
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.710812539987204e-05,
+      "loss": 3.8926,
+      "step": 6780
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.710599274898699e-05,
+      "loss": 4.0021,
+      "step": 6785
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.710386009810195e-05,
+      "loss": 3.9215,
+      "step": 6790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.710172744721689e-05,
+      "loss": 3.9291,
+      "step": 6795
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.709959479633184e-05,
+      "loss": 3.911,
+      "step": 6800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.70974621454468e-05,
+      "loss": 3.975,
+      "step": 6805
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.709532949456175e-05,
+      "loss": 4.0073,
+      "step": 6810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.70931968436767e-05,
+      "loss": 3.8968,
+      "step": 6815
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.709106419279165e-05,
+      "loss": 3.9345,
+      "step": 6820
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.708893154190659e-05,
+      "loss": 3.9894,
+      "step": 6825
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.708679889102154e-05,
+      "loss": 3.9378,
+      "step": 6830
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.708466624013649e-05,
+      "loss": 3.9816,
+      "step": 6835
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.708253358925144e-05,
+      "loss": 3.9267,
+      "step": 6840
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.70804009383664e-05,
+      "loss": 3.9295,
+      "step": 6845
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.707826828748135e-05,
+      "loss": 3.8927,
+      "step": 6850
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.70761356365963e-05,
+      "loss": 3.9844,
+      "step": 6855
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.707400298571125e-05,
+      "loss": 3.8569,
+      "step": 6860
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.70718703348262e-05,
+      "loss": 3.9604,
+      "step": 6865
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.706973768394114e-05,
+      "loss": 3.9782,
+      "step": 6870
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.70676050330561e-05,
+      "loss": 3.9195,
+      "step": 6875
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.706547238217104e-05,
+      "loss": 4.0089,
+      "step": 6880
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.706333973128599e-05,
+      "loss": 4.056,
+      "step": 6885
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.706120708040093e-05,
+      "loss": 4.0068,
+      "step": 6890
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.705907442951589e-05,
+      "loss": 4.0302,
+      "step": 6895
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.705694177863084e-05,
+      "loss": 3.9245,
+      "step": 6900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.70548091277458e-05,
+      "loss": 4.0525,
+      "step": 6905
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.705267647686074e-05,
+      "loss": 3.9153,
+      "step": 6910
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.70505438259757e-05,
+      "loss": 3.8977,
+      "step": 6915
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.704841117509065e-05,
+      "loss": 3.9582,
+      "step": 6920
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.704627852420559e-05,
+      "loss": 3.8941,
+      "step": 6925
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.704414587332054e-05,
+      "loss": 3.929,
+      "step": 6930
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.704201322243549e-05,
+      "loss": 3.8578,
+      "step": 6935
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.703988057155044e-05,
+      "loss": 3.9318,
+      "step": 6940
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.70377479206654e-05,
+      "loss": 4.081,
+      "step": 6945
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.703561526978035e-05,
+      "loss": 3.9084,
+      "step": 6950
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.703348261889529e-05,
+      "loss": 3.9184,
+      "step": 6955
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.703134996801025e-05,
+      "loss": 3.9983,
+      "step": 6960
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.702921731712519e-05,
+      "loss": 3.9593,
+      "step": 6965
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.702708466624014e-05,
+      "loss": 4.0072,
+      "step": 6970
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.702495201535508e-05,
+      "loss": 4.0001,
+      "step": 6975
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.702281936447004e-05,
+      "loss": 3.9286,
+      "step": 6980
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.702068671358499e-05,
+      "loss": 4.0657,
+      "step": 6985
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.701855406269993e-05,
+      "loss": 3.9592,
+      "step": 6990
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.701642141181489e-05,
+      "loss": 4.0018,
+      "step": 6995
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.701428876092984e-05,
+      "loss": 3.925,
+      "step": 7000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.70121561100448e-05,
+      "loss": 4.0501,
+      "step": 7005
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.701002345915974e-05,
+      "loss": 3.9154,
+      "step": 7010
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.70078908082747e-05,
+      "loss": 4.0242,
+      "step": 7015
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.700575815738964e-05,
+      "loss": 3.9354,
+      "step": 7020
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.700362550650459e-05,
+      "loss": 3.9424,
+      "step": 7025
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.700149285561953e-05,
+      "loss": 3.995,
+      "step": 7030
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.699936020473449e-05,
+      "loss": 3.999,
+      "step": 7035
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.699722755384944e-05,
+      "loss": 3.8797,
+      "step": 7040
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.69950949029644e-05,
+      "loss": 3.9442,
+      "step": 7045
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.699296225207935e-05,
+      "loss": 3.9112,
+      "step": 7050
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.699082960119429e-05,
+      "loss": 4.0144,
+      "step": 7055
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.698869695030925e-05,
+      "loss": 3.9404,
+      "step": 7060
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.698656429942419e-05,
+      "loss": 3.9207,
+      "step": 7065
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.698443164853914e-05,
+      "loss": 4.0416,
+      "step": 7070
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.698229899765408e-05,
+      "loss": 3.8205,
+      "step": 7075
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.698016634676904e-05,
+      "loss": 3.9883,
+      "step": 7080
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.697803369588398e-05,
+      "loss": 4.0117,
+      "step": 7085
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.697590104499893e-05,
+      "loss": 4.065,
+      "step": 7090
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.697376839411389e-05,
+      "loss": 3.9995,
+      "step": 7095
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.697163574322884e-05,
+      "loss": 4.017,
+      "step": 7100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.696950309234378e-05,
+      "loss": 3.9544,
+      "step": 7105
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.696737044145874e-05,
+      "loss": 3.959,
+      "step": 7110
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.69652377905737e-05,
+      "loss": 3.947,
+      "step": 7115
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.696310513968864e-05,
+      "loss": 3.8877,
+      "step": 7120
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.696097248880359e-05,
+      "loss": 3.9816,
+      "step": 7125
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.695883983791853e-05,
+      "loss": 3.9636,
+      "step": 7130
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.695670718703349e-05,
+      "loss": 4.034,
+      "step": 7135
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.695457453614844e-05,
+      "loss": 3.9534,
+      "step": 7140
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.69524418852634e-05,
+      "loss": 4.0422,
+      "step": 7145
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.695030923437834e-05,
+      "loss": 3.9076,
+      "step": 7150
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.694817658349329e-05,
+      "loss": 3.935,
+      "step": 7155
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.694604393260823e-05,
+      "loss": 3.9518,
+      "step": 7160
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.694391128172319e-05,
+      "loss": 3.9601,
+      "step": 7165
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.694177863083813e-05,
+      "loss": 3.9242,
+      "step": 7170
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.693964597995308e-05,
+      "loss": 4.0258,
+      "step": 7175
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.693751332906804e-05,
+      "loss": 3.9495,
+      "step": 7180
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.693538067818298e-05,
+      "loss": 3.9609,
+      "step": 7185
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.693324802729793e-05,
+      "loss": 3.9768,
+      "step": 7190
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.693111537641289e-05,
+      "loss": 3.9831,
+      "step": 7195
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.692898272552784e-05,
+      "loss": 3.9994,
+      "step": 7200
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.692685007464278e-05,
+      "loss": 4.0035,
+      "step": 7205
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.692471742375774e-05,
+      "loss": 3.9015,
+      "step": 7210
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.692258477287268e-05,
+      "loss": 3.9324,
+      "step": 7215
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.692045212198764e-05,
+      "loss": 4.0493,
+      "step": 7220
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.691831947110258e-05,
+      "loss": 3.8966,
+      "step": 7225
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.691618682021753e-05,
+      "loss": 4.096,
+      "step": 7230
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.691405416933249e-05,
+      "loss": 3.9641,
+      "step": 7235
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.691192151844744e-05,
+      "loss": 3.9442,
+      "step": 7240
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.69097888675624e-05,
+      "loss": 4.0565,
+      "step": 7245
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.690765621667734e-05,
+      "loss": 4.006,
+      "step": 7250
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.690552356579229e-05,
+      "loss": 3.8722,
+      "step": 7255
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.690339091490723e-05,
+      "loss": 4.0283,
+      "step": 7260
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.690125826402219e-05,
+      "loss": 3.8485,
+      "step": 7265
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689912561313713e-05,
+      "loss": 3.932,
+      "step": 7270
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689699296225208e-05,
+      "loss": 3.974,
+      "step": 7275
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689486031136702e-05,
+      "loss": 4.0064,
+      "step": 7280
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689272766048198e-05,
+      "loss": 4.0187,
+      "step": 7285
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689059500959693e-05,
+      "loss": 4.0893,
+      "step": 7290
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.688846235871189e-05,
+      "loss": 3.931,
+      "step": 7295
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.688632970782683e-05,
+      "loss": 4.0212,
+      "step": 7300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.688419705694178e-05,
+      "loss": 3.942,
+      "step": 7305
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.688206440605674e-05,
+      "loss": 3.9389,
+      "step": 7310
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.687993175517168e-05,
+      "loss": 3.9988,
+      "step": 7315
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.687779910428664e-05,
+      "loss": 3.9388,
+      "step": 7320
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.687566645340158e-05,
+      "loss": 3.8766,
+      "step": 7325
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.687353380251653e-05,
+      "loss": 3.9111,
+      "step": 7330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.687140115163149e-05,
+      "loss": 3.9832,
+      "step": 7335
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686926850074644e-05,
+      "loss": 3.9796,
+      "step": 7340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686713584986138e-05,
+      "loss": 3.9806,
+      "step": 7345
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686500319897634e-05,
+      "loss": 3.9975,
+      "step": 7350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686287054809128e-05,
+      "loss": 3.9825,
+      "step": 7355
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686073789720623e-05,
+      "loss": 4.0601,
+      "step": 7360
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685860524632117e-05,
+      "loss": 3.8426,
+      "step": 7365
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685647259543613e-05,
+      "loss": 3.8947,
+      "step": 7370
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685433994455108e-05,
+      "loss": 4.0457,
+      "step": 7375
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685220729366602e-05,
+      "loss": 4.033,
+      "step": 7380
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685007464278098e-05,
+      "loss": 4.0532,
+      "step": 7385
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.684794199189593e-05,
+      "loss": 4.0249,
+      "step": 7390
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.684580934101089e-05,
+      "loss": 3.8826,
+      "step": 7395
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.684367669012583e-05,
+      "loss": 3.9584,
+      "step": 7400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.684154403924078e-05,
+      "loss": 3.973,
+      "step": 7405
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.683941138835573e-05,
+      "loss": 3.9925,
+      "step": 7410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.683727873747068e-05,
+      "loss": 3.965,
+      "step": 7415
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.683514608658562e-05,
+      "loss": 3.9309,
+      "step": 7420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.683301343570058e-05,
+      "loss": 3.9501,
+      "step": 7425
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.683088078481553e-05,
+      "loss": 3.9286,
+      "step": 7430
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682874813393049e-05,
+      "loss": 3.9402,
+      "step": 7435
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682661548304544e-05,
+      "loss": 3.8433,
+      "step": 7440
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682448283216038e-05,
+      "loss": 3.8946,
+      "step": 7445
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682235018127534e-05,
+      "loss": 3.9796,
+      "step": 7450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682021753039028e-05,
+      "loss": 3.9952,
+      "step": 7455
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.681808487950523e-05,
+      "loss": 3.9109,
+      "step": 7460
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.681595222862017e-05,
+      "loss": 3.9025,
+      "step": 7465
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.681381957773513e-05,
+      "loss": 3.9738,
+      "step": 7470
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.681168692685007e-05,
+      "loss": 3.9674,
+      "step": 7475
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.680955427596502e-05,
+      "loss": 3.9367,
+      "step": 7480
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.680742162507998e-05,
+      "loss": 3.9728,
+      "step": 7485
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.680528897419493e-05,
+      "loss": 3.9568,
+      "step": 7490
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.680315632330988e-05,
+      "loss": 3.9422,
+      "step": 7495
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.680102367242483e-05,
+      "loss": 3.9362,
+      "step": 7500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.679889102153978e-05,
+      "loss": 3.9472,
+      "step": 7505
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.679675837065473e-05,
+      "loss": 3.9567,
+      "step": 7510
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.679462571976968e-05,
+      "loss": 4.069,
+      "step": 7515
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.679249306888462e-05,
+      "loss": 3.9777,
+      "step": 7520
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.679036041799958e-05,
+      "loss": 3.9746,
+      "step": 7525
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.678822776711453e-05,
+      "loss": 3.8673,
+      "step": 7530
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.678609511622949e-05,
+      "loss": 3.8358,
+      "step": 7535
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.678396246534443e-05,
+      "loss": 3.9602,
+      "step": 7540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.678182981445938e-05,
+      "loss": 3.9317,
+      "step": 7545
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677969716357432e-05,
+      "loss": 4.0068,
+      "step": 7550
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677756451268928e-05,
+      "loss": 3.9074,
+      "step": 7555
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677543186180422e-05,
+      "loss": 3.9604,
+      "step": 7560
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677329921091917e-05,
+      "loss": 3.9919,
+      "step": 7565
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677116656003413e-05,
+      "loss": 3.952,
+      "step": 7570
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.676903390914907e-05,
+      "loss": 3.9517,
+      "step": 7575
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.676690125826402e-05,
+      "loss": 4.0089,
+      "step": 7580
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.676476860737898e-05,
+      "loss": 3.9788,
+      "step": 7585
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.676263595649393e-05,
+      "loss": 3.9306,
+      "step": 7590
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.676050330560888e-05,
+      "loss": 3.9499,
+      "step": 7595
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.675837065472383e-05,
+      "loss": 3.9562,
+      "step": 7600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.675623800383877e-05,
+      "loss": 3.8718,
+      "step": 7605
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.675410535295373e-05,
+      "loss": 4.0027,
+      "step": 7610
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.675197270206867e-05,
+      "loss": 4.0228,
+      "step": 7615
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.674984005118362e-05,
+      "loss": 3.9901,
+      "step": 7620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.674770740029858e-05,
+      "loss": 4.001,
+      "step": 7625
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.674557474941353e-05,
+      "loss": 3.933,
+      "step": 7630
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.674344209852849e-05,
+      "loss": 4.038,
+      "step": 7635
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.674130944764343e-05,
+      "loss": 4.0128,
+      "step": 7640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.673917679675838e-05,
+      "loss": 3.9429,
+      "step": 7645
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.673704414587332e-05,
+      "loss": 4.0048,
+      "step": 7650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.673491149498828e-05,
+      "loss": 3.9505,
+      "step": 7655
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.673277884410322e-05,
+      "loss": 3.9534,
+      "step": 7660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.673064619321817e-05,
+      "loss": 3.9964,
+      "step": 7665
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.672851354233312e-05,
+      "loss": 4.0046,
+      "step": 7670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.672638089144807e-05,
+      "loss": 3.9452,
+      "step": 7675
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.672424824056302e-05,
+      "loss": 3.8981,
+      "step": 7680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.672211558967798e-05,
+      "loss": 3.9757,
+      "step": 7685
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671998293879292e-05,
+      "loss": 3.9287,
+      "step": 7690
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671785028790788e-05,
+      "loss": 3.7842,
+      "step": 7695
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671571763702283e-05,
+      "loss": 3.907,
+      "step": 7700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671358498613777e-05,
+      "loss": 3.9674,
+      "step": 7705
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671145233525273e-05,
+      "loss": 3.9987,
+      "step": 7710
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.670931968436767e-05,
+      "loss": 3.9149,
+      "step": 7715
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.670718703348262e-05,
+      "loss": 3.9172,
+      "step": 7720
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.670505438259758e-05,
+      "loss": 3.8956,
+      "step": 7725
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.670292173171253e-05,
+      "loss": 4.0116,
+      "step": 7730
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.670078908082747e-05,
+      "loss": 3.8976,
+      "step": 7735
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.669865642994243e-05,
+      "loss": 3.9813,
+      "step": 7740
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.669652377905737e-05,
+      "loss": 3.8436,
+      "step": 7745
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.669439112817232e-05,
+      "loss": 3.8233,
+      "step": 7750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.669225847728726e-05,
+      "loss": 3.9002,
+      "step": 7755
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.669012582640222e-05,
+      "loss": 3.8529,
+      "step": 7760
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.668799317551717e-05,
+      "loss": 3.868,
+      "step": 7765
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.668586052463212e-05,
+      "loss": 3.873,
+      "step": 7770
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.668372787374707e-05,
+      "loss": 3.93,
+      "step": 7775
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.668159522286202e-05,
+      "loss": 3.923,
+      "step": 7780
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.667946257197698e-05,
+      "loss": 4.0066,
+      "step": 7785
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.667732992109192e-05,
+      "loss": 3.9441,
+      "step": 7790
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.667519727020688e-05,
+      "loss": 3.9373,
+      "step": 7795
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.667306461932182e-05,
+      "loss": 4.0194,
+      "step": 7800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.667093196843677e-05,
+      "loss": 3.9457,
+      "step": 7805
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.666879931755171e-05,
+      "loss": 3.9548,
+      "step": 7810
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.666666666666667e-05,
+      "loss": 3.957,
+      "step": 7815
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.666453401578162e-05,
+      "loss": 3.9356,
+      "step": 7820
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.666240136489658e-05,
+      "loss": 3.8731,
+      "step": 7825
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.666026871401153e-05,
+      "loss": 3.9486,
+      "step": 7830
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.665813606312647e-05,
+      "loss": 4.0875,
+      "step": 7835
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.665600341224143e-05,
+      "loss": 3.9712,
+      "step": 7840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.665387076135637e-05,
+      "loss": 4.051,
+      "step": 7845
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.665173811047132e-05,
+      "loss": 3.955,
+      "step": 7850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.664960545958626e-05,
+      "loss": 4.1314,
+      "step": 7855
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.664747280870122e-05,
+      "loss": 3.9846,
+      "step": 7860
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.664534015781616e-05,
+      "loss": 3.8318,
+      "step": 7865
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.664320750693112e-05,
+      "loss": 3.9138,
+      "step": 7870
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.664107485604607e-05,
+      "loss": 3.8744,
+      "step": 7875
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663894220516102e-05,
+      "loss": 3.974,
+      "step": 7880
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663680955427597e-05,
+      "loss": 3.9319,
+      "step": 7885
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663467690339092e-05,
+      "loss": 3.9605,
+      "step": 7890
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663254425250588e-05,
+      "loss": 3.9157,
+      "step": 7895
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663041160162082e-05,
+      "loss": 3.9587,
+      "step": 7900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.662827895073577e-05,
+      "loss": 3.9459,
+      "step": 7905
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.662614629985071e-05,
+      "loss": 3.9782,
+      "step": 7910
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.662401364896567e-05,
+      "loss": 3.9681,
+      "step": 7915
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.662188099808062e-05,
+      "loss": 3.9228,
+      "step": 7920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.661974834719558e-05,
+      "loss": 3.9107,
+      "step": 7925
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.661761569631052e-05,
+      "loss": 3.9819,
+      "step": 7930
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.661548304542547e-05,
+      "loss": 3.9108,
+      "step": 7935
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.661335039454041e-05,
+      "loss": 3.9229,
+      "step": 7940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.661121774365537e-05,
+      "loss": 3.8541,
+      "step": 7945
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.660908509277031e-05,
+      "loss": 3.9974,
+      "step": 7950
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.660695244188526e-05,
+      "loss": 3.9336,
+      "step": 7955
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.660481979100022e-05,
+      "loss": 3.9691,
+      "step": 7960
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.660268714011516e-05,
+      "loss": 4.0531,
+      "step": 7965
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.660055448923012e-05,
+      "loss": 3.8823,
+      "step": 7970
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.659842183834507e-05,
+      "loss": 3.891,
+      "step": 7975
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.659628918746002e-05,
+      "loss": 3.9161,
+      "step": 7980
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.659415653657497e-05,
+      "loss": 3.8977,
+      "step": 7985
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.659202388568992e-05,
+      "loss": 3.97,
+      "step": 7990
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.658989123480486e-05,
+      "loss": 3.9983,
+      "step": 7995
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.658775858391982e-05,
+      "loss": 3.9206,
+      "step": 8000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.658562593303476e-05,
+      "loss": 3.858,
+      "step": 8005
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.658349328214971e-05,
+      "loss": 3.932,
+      "step": 8010
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.658136063126467e-05,
+      "loss": 3.9168,
+      "step": 8015
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657922798037962e-05,
+      "loss": 3.9086,
+      "step": 8020
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657709532949458e-05,
+      "loss": 3.8977,
+      "step": 8025
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657496267860952e-05,
+      "loss": 3.9408,
+      "step": 8030
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657283002772447e-05,
+      "loss": 3.974,
+      "step": 8035
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657069737683941e-05,
+      "loss": 3.9407,
+      "step": 8040
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.656856472595437e-05,
+      "loss": 3.9635,
+      "step": 8045
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.656643207506931e-05,
+      "loss": 3.9529,
+      "step": 8050
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.656429942418426e-05,
+      "loss": 3.8821,
+      "step": 8055
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.65621667732992e-05,
+      "loss": 3.9566,
+      "step": 8060
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.656003412241416e-05,
+      "loss": 3.9737,
+      "step": 8065
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.655790147152912e-05,
+      "loss": 3.9576,
+      "step": 8070
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.655576882064407e-05,
+      "loss": 4.0184,
+      "step": 8075
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.655363616975901e-05,
+      "loss": 3.9613,
+      "step": 8080
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.655150351887397e-05,
+      "loss": 4.0161,
+      "step": 8085
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.654937086798892e-05,
+      "loss": 3.9501,
+      "step": 8090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.654723821710386e-05,
+      "loss": 3.9022,
+      "step": 8095
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.654510556621882e-05,
+      "loss": 3.9753,
+      "step": 8100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.654297291533376e-05,
+      "loss": 3.8564,
+      "step": 8105
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.654084026444871e-05,
+      "loss": 3.9418,
+      "step": 8110
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.653870761356367e-05,
+      "loss": 3.8981,
+      "step": 8115
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.653657496267862e-05,
+      "loss": 3.8043,
+      "step": 8120
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.653444231179356e-05,
+      "loss": 3.8836,
+      "step": 8125
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.653230966090852e-05,
+      "loss": 4.028,
+      "step": 8130
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.653017701002346e-05,
+      "loss": 3.848,
+      "step": 8135
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.652804435913841e-05,
+      "loss": 3.9408,
+      "step": 8140
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.652591170825336e-05,
+      "loss": 3.9674,
+      "step": 8145
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.652377905736831e-05,
+      "loss": 3.9338,
+      "step": 8150
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.652164640648326e-05,
+      "loss": 3.9236,
+      "step": 8155
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.65195137555982e-05,
+      "loss": 3.8631,
+      "step": 8160
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.651738110471316e-05,
+      "loss": 4.0081,
+      "step": 8165
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.651524845382812e-05,
+      "loss": 3.9319,
+      "step": 8170
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.651311580294307e-05,
+      "loss": 3.9249,
+      "step": 8175
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.651098315205801e-05,
+      "loss": 3.9253,
+      "step": 8180
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.650885050117297e-05,
+      "loss": 3.9863,
+      "step": 8185
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.650671785028791e-05,
+      "loss": 3.9515,
+      "step": 8190
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.650458519940286e-05,
+      "loss": 3.9285,
+      "step": 8195
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.65024525485178e-05,
+      "loss": 3.9682,
+      "step": 8200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.650031989763276e-05,
+      "loss": 3.9667,
+      "step": 8205
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.649818724674771e-05,
+      "loss": 4.0204,
+      "step": 8210
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.649605459586267e-05,
+      "loss": 3.9949,
+      "step": 8215
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.649392194497762e-05,
+      "loss": 3.9062,
+      "step": 8220
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.649178929409256e-05,
+      "loss": 3.9013,
+      "step": 8225
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.648965664320752e-05,
+      "loss": 3.9498,
+      "step": 8230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.648752399232246e-05,
+      "loss": 4.0275,
+      "step": 8235
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.648539134143741e-05,
+      "loss": 3.9451,
+      "step": 8240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.648325869055236e-05,
+      "loss": 3.9771,
+      "step": 8245
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.648112603966731e-05,
+      "loss": 3.9929,
+      "step": 8250
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.647899338878225e-05,
+      "loss": 4.0019,
+      "step": 8255
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.64768607378972e-05,
+      "loss": 3.9003,
+      "step": 8260
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.647472808701216e-05,
+      "loss": 4.0787,
+      "step": 8265
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.647259543612712e-05,
+      "loss": 3.8502,
+      "step": 8270
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.647046278524206e-05,
+      "loss": 3.8723,
+      "step": 8275
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.646833013435701e-05,
+      "loss": 3.8762,
+      "step": 8280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.646619748347197e-05,
+      "loss": 3.982,
+      "step": 8285
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.646406483258691e-05,
+      "loss": 3.9126,
+      "step": 8290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.646193218170186e-05,
+      "loss": 4.0413,
+      "step": 8295
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.64597995308168e-05,
+      "loss": 3.9974,
+      "step": 8300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.645766687993176e-05,
+      "loss": 3.9017,
+      "step": 8305
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.645553422904671e-05,
+      "loss": 4.0687,
+      "step": 8310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.645340157816167e-05,
+      "loss": 3.9069,
+      "step": 8315
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.645126892727661e-05,
+      "loss": 3.888,
+      "step": 8320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.644913627639156e-05,
+      "loss": 3.9087,
+      "step": 8325
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.64470036255065e-05,
+      "loss": 3.9771,
+      "step": 8330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.644487097462146e-05,
+      "loss": 3.9748,
+      "step": 8335
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.64427383237364e-05,
+      "loss": 4.0026,
+      "step": 8340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.644060567285136e-05,
+      "loss": 3.895,
+      "step": 8345
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.643847302196631e-05,
+      "loss": 4.012,
+      "step": 8350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.643634037108125e-05,
+      "loss": 3.87,
+      "step": 8355
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.64342077201962e-05,
+      "loss": 4.0043,
+      "step": 8360
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.643207506931116e-05,
+      "loss": 4.0,
+      "step": 8365
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.642994241842612e-05,
+      "loss": 3.9676,
+      "step": 8370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.642780976754106e-05,
+      "loss": 3.9396,
+      "step": 8375
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.642567711665601e-05,
+      "loss": 3.9793,
+      "step": 8380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.642354446577095e-05,
+      "loss": 3.9329,
+      "step": 8385
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.642141181488591e-05,
+      "loss": 4.0025,
+      "step": 8390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.641927916400085e-05,
+      "loss": 3.9627,
+      "step": 8395
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.64171465131158e-05,
+      "loss": 3.936,
+      "step": 8400
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.641501386223076e-05,
+      "loss": 3.9795,
+      "step": 8405
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.641288121134571e-05,
+      "loss": 3.9362,
+      "step": 8410
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.641074856046067e-05,
+      "loss": 3.931,
+      "step": 8415
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.640861590957561e-05,
+      "loss": 3.904,
+      "step": 8420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.640648325869056e-05,
+      "loss": 3.9252,
+      "step": 8425
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.64043506078055e-05,
+      "loss": 3.794,
+      "step": 8430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.640221795692046e-05,
+      "loss": 3.8642,
+      "step": 8435
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.64000853060354e-05,
+      "loss": 3.9434,
+      "step": 8440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.639795265515036e-05,
+      "loss": 3.9625,
+      "step": 8445
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.63958200042653e-05,
+      "loss": 3.9158,
+      "step": 8450
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.639368735338025e-05,
+      "loss": 3.9622,
+      "step": 8455
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.63915547024952e-05,
+      "loss": 3.8525,
+      "step": 8460
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.638942205161016e-05,
+      "loss": 3.9034,
+      "step": 8465
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.63872894007251e-05,
+      "loss": 3.9927,
+      "step": 8470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.638515674984006e-05,
+      "loss": 3.941,
+      "step": 8475
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.638302409895501e-05,
+      "loss": 3.9747,
+      "step": 8480
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.638089144806995e-05,
+      "loss": 4.0354,
+      "step": 8485
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637875879718491e-05,
+      "loss": 3.9309,
+      "step": 8490
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637662614629985e-05,
+      "loss": 3.8677,
+      "step": 8495
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.63744934954148e-05,
+      "loss": 3.9659,
+      "step": 8500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637236084452976e-05,
+      "loss": 3.9476,
+      "step": 8505
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637022819364471e-05,
+      "loss": 3.9306,
+      "step": 8510
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.636809554275965e-05,
+      "loss": 3.9389,
+      "step": 8515
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.636596289187461e-05,
+      "loss": 3.9782,
+      "step": 8520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.636383024098955e-05,
+      "loss": 3.9809,
+      "step": 8525
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.63616975901045e-05,
+      "loss": 3.867,
+      "step": 8530
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.635956493921946e-05,
+      "loss": 4.043,
+      "step": 8535
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.63574322883344e-05,
+      "loss": 3.9205,
+      "step": 8540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.635529963744936e-05,
+      "loss": 3.916,
+      "step": 8545
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.63531669865643e-05,
+      "loss": 3.9487,
+      "step": 8550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.635103433567925e-05,
+      "loss": 3.9018,
+      "step": 8555
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.63489016847942e-05,
+      "loss": 3.8664,
+      "step": 8560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.634676903390916e-05,
+      "loss": 3.9274,
+      "step": 8565
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.63446363830241e-05,
+      "loss": 4.0504,
+      "step": 8570
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.634250373213906e-05,
+      "loss": 3.9087,
+      "step": 8575
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.6340371081254e-05,
+      "loss": 3.9666,
+      "step": 8580
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.633823843036895e-05,
+      "loss": 3.9854,
+      "step": 8585
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.63361057794839e-05,
+      "loss": 3.8774,
+      "step": 8590
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.633397312859885e-05,
+      "loss": 4.0348,
+      "step": 8595
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.63318404777138e-05,
+      "loss": 3.9192,
+      "step": 8600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.632970782682876e-05,
+      "loss": 3.9169,
+      "step": 8605
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.632757517594371e-05,
+      "loss": 3.9721,
+      "step": 8610
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.632544252505865e-05,
+      "loss": 3.9537,
+      "step": 8615
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.632330987417361e-05,
+      "loss": 3.997,
+      "step": 8620
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.632117722328855e-05,
+      "loss": 3.8441,
+      "step": 8625
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.63190445724035e-05,
+      "loss": 3.8964,
+      "step": 8630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.631691192151845e-05,
+      "loss": 3.91,
+      "step": 8635
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.63147792706334e-05,
+      "loss": 3.9637,
+      "step": 8640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.631264661974834e-05,
+      "loss": 3.816,
+      "step": 8645
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.63105139688633e-05,
+      "loss": 3.991,
+      "step": 8650
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.630838131797825e-05,
+      "loss": 3.8514,
+      "step": 8655
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.63062486670932e-05,
+      "loss": 3.9244,
+      "step": 8660
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.630411601620816e-05,
+      "loss": 3.959,
+      "step": 8665
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.63019833653231e-05,
+      "loss": 3.8843,
+      "step": 8670
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.629985071443806e-05,
+      "loss": 3.9632,
+      "step": 8675
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.6297718063553e-05,
+      "loss": 3.9585,
+      "step": 8680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.629558541266795e-05,
+      "loss": 3.8212,
+      "step": 8685
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.62934527617829e-05,
+      "loss": 3.8204,
+      "step": 8690
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.629132011089785e-05,
+      "loss": 3.928,
+      "step": 8695
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.62891874600128e-05,
+      "loss": 3.9056,
+      "step": 8700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.628705480912776e-05,
+      "loss": 3.8705,
+      "step": 8705
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.62849221582427e-05,
+      "loss": 3.942,
+      "step": 8710
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.628278950735765e-05,
+      "loss": 3.8656,
+      "step": 8715
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.62806568564726e-05,
+      "loss": 3.8137,
+      "step": 8720
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.627852420558755e-05,
+      "loss": 3.8635,
+      "step": 8725
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.62763915547025e-05,
+      "loss": 3.9186,
+      "step": 8730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.627425890381745e-05,
+      "loss": 3.8304,
+      "step": 8735
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.62721262529324e-05,
+      "loss": 3.9784,
+      "step": 8740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.626999360204734e-05,
+      "loss": 3.8235,
+      "step": 8745
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.62678609511623e-05,
+      "loss": 3.9063,
+      "step": 8750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.626572830027725e-05,
+      "loss": 3.8882,
+      "step": 8755
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.62635956493922e-05,
+      "loss": 3.8283,
+      "step": 8760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.626146299850715e-05,
+      "loss": 3.8209,
+      "step": 8765
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.62593303476221e-05,
+      "loss": 3.8694,
+      "step": 8770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.625719769673704e-05,
+      "loss": 3.9335,
+      "step": 8775
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.6255065045852e-05,
+      "loss": 3.8924,
+      "step": 8780
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.625293239496694e-05,
+      "loss": 3.9002,
+      "step": 8785
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.62507997440819e-05,
+      "loss": 3.9546,
+      "step": 8790
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.624866709319685e-05,
+      "loss": 3.9759,
+      "step": 8795
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.62465344423118e-05,
+      "loss": 3.9054,
+      "step": 8800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.624440179142676e-05,
+      "loss": 3.9971,
+      "step": 8805
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.62422691405417e-05,
+      "loss": 3.9312,
+      "step": 8810
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.624013648965665e-05,
+      "loss": 3.9775,
+      "step": 8815
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.62380038387716e-05,
+      "loss": 3.8942,
+      "step": 8820
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.623587118788655e-05,
+      "loss": 3.9469,
+      "step": 8825
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.623373853700149e-05,
+      "loss": 3.9949,
+      "step": 8830
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.623160588611645e-05,
+      "loss": 3.8898,
+      "step": 8835
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.622947323523139e-05,
+      "loss": 3.9294,
+      "step": 8840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.622734058434634e-05,
+      "loss": 3.9123,
+      "step": 8845
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.62252079334613e-05,
+      "loss": 3.9071,
+      "step": 8850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.622307528257625e-05,
+      "loss": 3.9909,
+      "step": 8855
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.62209426316912e-05,
+      "loss": 3.9856,
+      "step": 8860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.621880998080615e-05,
+      "loss": 3.9783,
+      "step": 8865
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.62166773299211e-05,
+      "loss": 3.89,
+      "step": 8870
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.621454467903604e-05,
+      "loss": 3.938,
+      "step": 8875
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.6212412028151e-05,
+      "loss": 3.8597,
+      "step": 8880
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.621027937726594e-05,
+      "loss": 3.9513,
+      "step": 8885
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.62081467263809e-05,
+      "loss": 3.8398,
+      "step": 8890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.620601407549585e-05,
+      "loss": 3.9375,
+      "step": 8895
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.62038814246108e-05,
+      "loss": 3.8109,
+      "step": 8900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.620174877372574e-05,
+      "loss": 3.9151,
+      "step": 8905
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.61996161228407e-05,
+      "loss": 3.865,
+      "step": 8910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.619748347195564e-05,
+      "loss": 4.0337,
+      "step": 8915
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.61953508210706e-05,
+      "loss": 3.8709,
+      "step": 8920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.619321817018555e-05,
+      "loss": 3.8984,
+      "step": 8925
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.619108551930049e-05,
+      "loss": 3.9675,
+      "step": 8930
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.618895286841545e-05,
+      "loss": 3.8844,
+      "step": 8935
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.618682021753039e-05,
+      "loss": 3.9666,
+      "step": 8940
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.618468756664534e-05,
+      "loss": 3.9984,
+      "step": 8945
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.61825549157603e-05,
+      "loss": 3.8754,
+      "step": 8950
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.618042226487525e-05,
+      "loss": 3.8035,
+      "step": 8955
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.617828961399019e-05,
+      "loss": 3.8567,
+      "step": 8960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.617615696310515e-05,
+      "loss": 4.0156,
+      "step": 8965
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.617402431222009e-05,
+      "loss": 3.9287,
+      "step": 8970
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.617189166133504e-05,
+      "loss": 3.8815,
+      "step": 8975
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.616975901044998e-05,
+      "loss": 4.0201,
+      "step": 8980
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.616762635956494e-05,
+      "loss": 3.9027,
+      "step": 8985
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.61654937086799e-05,
+      "loss": 3.8751,
+      "step": 8990
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.616336105779485e-05,
+      "loss": 3.807,
+      "step": 8995
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.61612284069098e-05,
+      "loss": 3.9068,
+      "step": 9000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.615909575602474e-05,
+      "loss": 3.9204,
+      "step": 9005
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.61569631051397e-05,
+      "loss": 3.8568,
+      "step": 9010
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.615483045425464e-05,
+      "loss": 3.9047,
+      "step": 9015
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.61526978033696e-05,
+      "loss": 3.9855,
+      "step": 9020
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.615056515248454e-05,
+      "loss": 3.9231,
+      "step": 9025
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.614843250159949e-05,
+      "loss": 4.0087,
+      "step": 9030
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.614629985071443e-05,
+      "loss": 4.0523,
+      "step": 9035
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.614416719982939e-05,
+      "loss": 3.9145,
+      "step": 9040
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.614203454894434e-05,
+      "loss": 3.7717,
+      "step": 9045
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.61399018980593e-05,
+      "loss": 3.7863,
+      "step": 9050
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.613776924717425e-05,
+      "loss": 3.9593,
+      "step": 9055
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.613563659628919e-05,
+      "loss": 3.9541,
+      "step": 9060
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.613350394540415e-05,
+      "loss": 3.9072,
+      "step": 9065
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.613137129451909e-05,
+      "loss": 3.8733,
+      "step": 9070
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.612923864363404e-05,
+      "loss": 3.9044,
+      "step": 9075
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.612710599274898e-05,
+      "loss": 3.9704,
+      "step": 9080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.612497334186394e-05,
+      "loss": 3.8255,
+      "step": 9085
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.61228406909789e-05,
+      "loss": 3.862,
+      "step": 9090
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.612070804009385e-05,
+      "loss": 3.9216,
+      "step": 9095
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.611857538920879e-05,
+      "loss": 3.8865,
+      "step": 9100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.611644273832374e-05,
+      "loss": 3.9935,
+      "step": 9105
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.611431008743869e-05,
+      "loss": 3.9115,
+      "step": 9110
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.611217743655364e-05,
+      "loss": 3.8866,
+      "step": 9115
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.61100447856686e-05,
+      "loss": 3.888,
+      "step": 9120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.610791213478354e-05,
+      "loss": 3.9667,
+      "step": 9125
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.610577948389849e-05,
+      "loss": 3.9008,
+      "step": 9130
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.610364683301343e-05,
+      "loss": 3.9049,
+      "step": 9135
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.610151418212839e-05,
+      "loss": 3.9664,
+      "step": 9140
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.609938153124334e-05,
+      "loss": 3.9578,
+      "step": 9145
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.60972488803583e-05,
+      "loss": 3.9024,
+      "step": 9150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.609511622947324e-05,
+      "loss": 3.9579,
+      "step": 9155
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.609298357858819e-05,
+      "loss": 3.9001,
+      "step": 9160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.609085092770313e-05,
+      "loss": 3.8728,
+      "step": 9165
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.608871827681809e-05,
+      "loss": 3.9192,
+      "step": 9170
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.608658562593303e-05,
+      "loss": 3.9924,
+      "step": 9175
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.608445297504798e-05,
+      "loss": 3.9081,
+      "step": 9180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.608232032416294e-05,
+      "loss": 3.9442,
+      "step": 9185
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.60801876732779e-05,
+      "loss": 3.921,
+      "step": 9190
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.607805502239285e-05,
+      "loss": 3.9555,
+      "step": 9195
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.607592237150779e-05,
+      "loss": 3.9979,
+      "step": 9200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.607378972062274e-05,
+      "loss": 3.8777,
+      "step": 9205
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.607165706973769e-05,
+      "loss": 3.8775,
+      "step": 9210
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.606952441885264e-05,
+      "loss": 3.9365,
+      "step": 9215
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.606739176796758e-05,
+      "loss": 3.9906,
+      "step": 9220
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.606525911708254e-05,
+      "loss": 3.8708,
+      "step": 9225
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.606312646619748e-05,
+      "loss": 3.9532,
+      "step": 9230
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.606099381531243e-05,
+      "loss": 3.9191,
+      "step": 9235
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.605886116442739e-05,
+      "loss": 3.9653,
+      "step": 9240
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.605672851354234e-05,
+      "loss": 3.9414,
+      "step": 9245
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.60545958626573e-05,
+      "loss": 4.0171,
+      "step": 9250
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.605246321177224e-05,
+      "loss": 3.8316,
+      "step": 9255
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.605033056088719e-05,
+      "loss": 4.0025,
+      "step": 9260
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.604819791000213e-05,
+      "loss": 3.9262,
+      "step": 9265
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.604606525911709e-05,
+      "loss": 3.9224,
+      "step": 9270
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.604393260823203e-05,
+      "loss": 3.9596,
+      "step": 9275
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.604179995734698e-05,
+      "loss": 3.9289,
+      "step": 9280
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.603966730646194e-05,
+      "loss": 3.8967,
+      "step": 9285
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.60375346555769e-05,
+      "loss": 3.8816,
+      "step": 9290
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.603540200469183e-05,
+      "loss": 3.7878,
+      "step": 9295
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.603326935380679e-05,
+      "loss": 3.8872,
+      "step": 9300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.603113670292173e-05,
+      "loss": 3.8306,
+      "step": 9305
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.602900405203669e-05,
+      "loss": 3.7977,
+      "step": 9310
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.602687140115164e-05,
+      "loss": 3.9371,
+      "step": 9315
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.602473875026658e-05,
+      "loss": 3.9401,
+      "step": 9320
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.602260609938154e-05,
+      "loss": 3.88,
+      "step": 9325
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.602047344849648e-05,
+      "loss": 3.8741,
+      "step": 9330
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.601834079761145e-05,
+      "loss": 4.0231,
+      "step": 9335
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.601620814672639e-05,
+      "loss": 3.7976,
+      "step": 9340
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.601407549584134e-05,
+      "loss": 3.9515,
+      "step": 9345
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.601194284495628e-05,
+      "loss": 3.9097,
+      "step": 9350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.600981019407124e-05,
+      "loss": 3.9143,
+      "step": 9355
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.600767754318618e-05,
+      "loss": 3.9404,
+      "step": 9360
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.600554489230113e-05,
+      "loss": 3.9616,
+      "step": 9365
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.600341224141607e-05,
+      "loss": 4.005,
+      "step": 9370
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.600127959053103e-05,
+      "loss": 3.9032,
+      "step": 9375
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.599914693964598e-05,
+      "loss": 3.9546,
+      "step": 9380
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.599701428876094e-05,
+      "loss": 3.9168,
+      "step": 9385
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.59948816378759e-05,
+      "loss": 3.9348,
+      "step": 9390
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.599274898699084e-05,
+      "loss": 3.975,
+      "step": 9395
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.599061633610579e-05,
+      "loss": 3.9352,
+      "step": 9400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.598848368522073e-05,
+      "loss": 3.9051,
+      "step": 9405
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.598635103433569e-05,
+      "loss": 3.8676,
+      "step": 9410
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.598421838345063e-05,
+      "loss": 3.8709,
+      "step": 9415
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.598208573256558e-05,
+      "loss": 3.8946,
+      "step": 9420
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.597995308168052e-05,
+      "loss": 3.8885,
+      "step": 9425
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.597782043079548e-05,
+      "loss": 3.9497,
+      "step": 9430
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.597568777991043e-05,
+      "loss": 3.797,
+      "step": 9435
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.597355512902539e-05,
+      "loss": 3.9236,
+      "step": 9440
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.597142247814034e-05,
+      "loss": 3.8859,
+      "step": 9445
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.596928982725528e-05,
+      "loss": 3.8632,
+      "step": 9450
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.596715717637024e-05,
+      "loss": 3.894,
+      "step": 9455
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.596502452548518e-05,
+      "loss": 3.8814,
+      "step": 9460
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.596289187460013e-05,
+      "loss": 3.934,
+      "step": 9465
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.596075922371507e-05,
+      "loss": 3.8968,
+      "step": 9470
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595862657283003e-05,
+      "loss": 3.9153,
+      "step": 9475
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595649392194498e-05,
+      "loss": 3.8999,
+      "step": 9480
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595436127105994e-05,
+      "loss": 3.8634,
+      "step": 9485
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595222862017488e-05,
+      "loss": 3.7982,
+      "step": 9490
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595009596928984e-05,
+      "loss": 3.8526,
+      "step": 9495
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.594796331840478e-05,
+      "loss": 3.8855,
+      "step": 9500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.594583066751973e-05,
+      "loss": 3.9068,
+      "step": 9505
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.594369801663469e-05,
+      "loss": 3.9052,
+      "step": 9510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.594156536574963e-05,
+      "loss": 3.9008,
+      "step": 9515
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.593943271486458e-05,
+      "loss": 3.887,
+      "step": 9520
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.593730006397952e-05,
+      "loss": 3.7917,
+      "step": 9525
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.593516741309449e-05,
+      "loss": 3.9603,
+      "step": 9530
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.593303476220943e-05,
+      "loss": 3.9165,
+      "step": 9535
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.593090211132439e-05,
+      "loss": 3.8178,
+      "step": 9540
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.592876946043933e-05,
+      "loss": 3.9096,
+      "step": 9545
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.592663680955428e-05,
+      "loss": 3.8815,
+      "step": 9550
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.592450415866922e-05,
+      "loss": 3.8176,
+      "step": 9555
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.592237150778418e-05,
+      "loss": 3.8058,
+      "step": 9560
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.592023885689912e-05,
+      "loss": 3.8784,
+      "step": 9565
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.591810620601407e-05,
+      "loss": 3.8491,
+      "step": 9570
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.591597355512903e-05,
+      "loss": 3.9405,
+      "step": 9575
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.591384090424398e-05,
+      "loss": 3.9151,
+      "step": 9580
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.591170825335894e-05,
+      "loss": 3.9383,
+      "step": 9585
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.590957560247388e-05,
+      "loss": 3.7983,
+      "step": 9590
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590744295158884e-05,
+      "loss": 3.9112,
+      "step": 9595
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590531030070378e-05,
+      "loss": 3.8687,
+      "step": 9600
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590317764981873e-05,
+      "loss": 3.9013,
+      "step": 9605
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590104499893367e-05,
+      "loss": 3.915,
+      "step": 9610
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.589891234804863e-05,
+      "loss": 3.9355,
+      "step": 9615
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.589677969716357e-05,
+      "loss": 3.8931,
+      "step": 9620
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.589464704627852e-05,
+      "loss": 3.8911,
+      "step": 9625
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.589251439539348e-05,
+      "loss": 3.9141,
+      "step": 9630
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.589038174450843e-05,
+      "loss": 3.8991,
+      "step": 9635
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.588824909362339e-05,
+      "loss": 3.9115,
+      "step": 9640
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.588611644273833e-05,
+      "loss": 3.9708,
+      "step": 9645
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.588398379185328e-05,
+      "loss": 3.9296,
+      "step": 9650
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.588185114096822e-05,
+      "loss": 3.8931,
+      "step": 9655
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.587971849008318e-05,
+      "loss": 3.8585,
+      "step": 9660
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.587758583919812e-05,
+      "loss": 3.8648,
+      "step": 9665
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.587545318831307e-05,
+      "loss": 3.8755,
+      "step": 9670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.587332053742803e-05,
+      "loss": 3.9061,
+      "step": 9675
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.587118788654298e-05,
+      "loss": 3.931,
+      "step": 9680
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.586905523565793e-05,
+      "loss": 3.8748,
+      "step": 9685
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.586692258477288e-05,
+      "loss": 3.9259,
+      "step": 9690
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.586478993388782e-05,
+      "loss": 3.8767,
+      "step": 9695
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.586265728300278e-05,
+      "loss": 3.8803,
+      "step": 9700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.586052463211773e-05,
+      "loss": 3.905,
+      "step": 9705
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.585839198123267e-05,
+      "loss": 3.8924,
+      "step": 9710
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.585625933034763e-05,
+      "loss": 3.873,
+      "step": 9715
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.585412667946257e-05,
+      "loss": 3.8622,
+      "step": 9720
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.585199402857754e-05,
+      "loss": 3.8733,
+      "step": 9725
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.584986137769248e-05,
+      "loss": 3.9065,
+      "step": 9730
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.584772872680743e-05,
+      "loss": 4.0425,
+      "step": 9735
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.584559607592237e-05,
+      "loss": 3.8029,
+      "step": 9740
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.584346342503733e-05,
+      "loss": 3.9621,
+      "step": 9745
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.584133077415227e-05,
+      "loss": 3.917,
+      "step": 9750
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.583919812326722e-05,
+      "loss": 3.8314,
+      "step": 9755
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.583706547238217e-05,
+      "loss": 3.7895,
+      "step": 9760
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.583493282149712e-05,
+      "loss": 3.8788,
+      "step": 9765
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.583280017061207e-05,
+      "loss": 3.9547,
+      "step": 9770
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.583066751972703e-05,
+      "loss": 3.8664,
+      "step": 9775
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.582853486884198e-05,
+      "loss": 3.863,
+      "step": 9780
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.582640221795693e-05,
+      "loss": 3.8867,
+      "step": 9785
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.582426956707188e-05,
+      "loss": 3.9876,
+      "step": 9790
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.582213691618682e-05,
+      "loss": 3.9828,
+      "step": 9795
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.582000426530178e-05,
+      "loss": 3.8856,
+      "step": 9800
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.581787161441672e-05,
+      "loss": 3.8726,
+      "step": 9805
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.581573896353167e-05,
+      "loss": 3.8798,
+      "step": 9810
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.581360631264661e-05,
+      "loss": 3.8873,
+      "step": 9815
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.581147366176157e-05,
+      "loss": 3.8703,
+      "step": 9820
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.580934101087652e-05,
+      "loss": 3.8842,
+      "step": 9825
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.580720835999148e-05,
+      "loss": 3.769,
+      "step": 9830
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.580507570910643e-05,
+      "loss": 3.9277,
+      "step": 9835
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.580294305822137e-05,
+      "loss": 3.8941,
+      "step": 9840
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.580081040733633e-05,
+      "loss": 3.9055,
+      "step": 9845
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.579867775645127e-05,
+      "loss": 3.8511,
+      "step": 9850
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.579654510556622e-05,
+      "loss": 3.8535,
+      "step": 9855
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.579441245468117e-05,
+      "loss": 3.9194,
+      "step": 9860
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.579227980379612e-05,
+      "loss": 4.0642,
+      "step": 9865
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.579014715291107e-05,
+      "loss": 3.7907,
+      "step": 9870
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.578801450202603e-05,
+      "loss": 3.8971,
+      "step": 9875
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.578588185114097e-05,
+      "loss": 3.9967,
+      "step": 9880
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.578374920025593e-05,
+      "loss": 3.8269,
+      "step": 9885
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.578161654937087e-05,
+      "loss": 3.7898,
+      "step": 9890
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.577948389848582e-05,
+      "loss": 3.9452,
+      "step": 9895
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.577735124760078e-05,
+      "loss": 3.8991,
+      "step": 9900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.577521859671572e-05,
+      "loss": 3.9386,
+      "step": 9905
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.577308594583067e-05,
+      "loss": 3.8612,
+      "step": 9910
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.577095329494561e-05,
+      "loss": 3.9359,
+      "step": 9915
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.576882064406058e-05,
+      "loss": 3.9041,
+      "step": 9920
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.576668799317552e-05,
+      "loss": 3.9987,
+      "step": 9925
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.576455534229048e-05,
+      "loss": 3.8928,
+      "step": 9930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.576242269140542e-05,
+      "loss": 3.9351,
+      "step": 9935
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.576029004052037e-05,
+      "loss": 3.8734,
+      "step": 9940
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.575815738963531e-05,
+      "loss": 3.9403,
+      "step": 9945
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.575602473875027e-05,
+      "loss": 3.9082,
+      "step": 9950
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.575389208786521e-05,
+      "loss": 3.9097,
+      "step": 9955
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.575175943698017e-05,
+      "loss": 3.93,
+      "step": 9960
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.574962678609512e-05,
+      "loss": 3.9413,
+      "step": 9965
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.574749413521007e-05,
+      "loss": 3.8804,
+      "step": 9970
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.574536148432503e-05,
+      "loss": 3.8673,
+      "step": 9975
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.574322883343997e-05,
+      "loss": 3.8399,
+      "step": 9980
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.574109618255493e-05,
+      "loss": 3.8549,
+      "step": 9985
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.573896353166987e-05,
+      "loss": 3.8775,
+      "step": 9990
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.573683088078482e-05,
+      "loss": 3.865,
+      "step": 9995
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.573469822989976e-05,
+      "loss": 3.8723,
+      "step": 10000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.573256557901472e-05,
+      "loss": 3.8955,
+      "step": 10005
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.573043292812966e-05,
+      "loss": 3.9485,
+      "step": 10010
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.572830027724463e-05,
+      "loss": 3.8632,
+      "step": 10015
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.572616762635957e-05,
+      "loss": 3.9496,
+      "step": 10020
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.572403497547452e-05,
+      "loss": 3.9024,
+      "step": 10025
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.572190232458948e-05,
+      "loss": 3.8518,
+      "step": 10030
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.571976967370442e-05,
+      "loss": 3.901,
+      "step": 10035
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.571763702281937e-05,
+      "loss": 3.8741,
+      "step": 10040
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.571550437193431e-05,
+      "loss": 3.8388,
+      "step": 10045
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.571337172104927e-05,
+      "loss": 3.8321,
+      "step": 10050
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.571123907016421e-05,
+      "loss": 3.9299,
+      "step": 10055
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.570910641927917e-05,
+      "loss": 3.8561,
+      "step": 10060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.570697376839412e-05,
+      "loss": 3.8446,
+      "step": 10065
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.570484111750907e-05,
+      "loss": 3.8178,
+      "step": 10070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.570270846662402e-05,
+      "loss": 3.9552,
+      "step": 10075
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.570057581573897e-05,
+      "loss": 3.7841,
+      "step": 10080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.569844316485391e-05,
+      "loss": 3.8744,
+      "step": 10085
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.569631051396887e-05,
+      "loss": 3.9294,
+      "step": 10090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.569417786308382e-05,
+      "loss": 3.7963,
+      "step": 10095
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.569204521219876e-05,
+      "loss": 3.9618,
+      "step": 10100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.568991256131372e-05,
+      "loss": 3.9147,
+      "step": 10105
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.568777991042866e-05,
+      "loss": 3.8262,
+      "step": 10110
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.568564725954363e-05,
+      "loss": 3.8093,
+      "step": 10115
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.568351460865857e-05,
+      "loss": 3.9012,
+      "step": 10120
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.568138195777352e-05,
+      "loss": 3.9055,
+      "step": 10125
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.567924930688846e-05,
+      "loss": 3.8433,
+      "step": 10130
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.567711665600342e-05,
+      "loss": 3.8674,
+      "step": 10135
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.567498400511836e-05,
+      "loss": 4.0064,
+      "step": 10140
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.567285135423331e-05,
+      "loss": 3.8263,
+      "step": 10145
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.567071870334826e-05,
+      "loss": 3.8567,
+      "step": 10150
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566858605246321e-05,
+      "loss": 3.882,
+      "step": 10155
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566645340157817e-05,
+      "loss": 3.8634,
+      "step": 10160
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566432075069312e-05,
+      "loss": 4.0028,
+      "step": 10165
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566218809980807e-05,
+      "loss": 3.8764,
+      "step": 10170
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566005544892302e-05,
+      "loss": 3.8509,
+      "step": 10175
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.565792279803797e-05,
+      "loss": 3.9297,
+      "step": 10180
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.565579014715291e-05,
+      "loss": 3.885,
+      "step": 10185
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.565365749626787e-05,
+      "loss": 3.8632,
+      "step": 10190
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.565152484538281e-05,
+      "loss": 3.8047,
+      "step": 10195
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.564939219449776e-05,
+      "loss": 3.9207,
+      "step": 10200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.56472595436127e-05,
+      "loss": 3.842,
+      "step": 10205
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.564512689272767e-05,
+      "loss": 3.8528,
+      "step": 10210
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.564299424184261e-05,
+      "loss": 3.9084,
+      "step": 10215
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.564086159095757e-05,
+      "loss": 3.9161,
+      "step": 10220
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.563872894007252e-05,
+      "loss": 3.8518,
+      "step": 10225
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.563659628918746e-05,
+      "loss": 3.7427,
+      "step": 10230
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.563446363830242e-05,
+      "loss": 3.9175,
+      "step": 10235
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.563233098741736e-05,
+      "loss": 3.8272,
+      "step": 10240
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.563019833653231e-05,
+      "loss": 3.9403,
+      "step": 10245
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.562806568564726e-05,
+      "loss": 3.7691,
+      "step": 10250
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.562593303476221e-05,
+      "loss": 3.856,
+      "step": 10255
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.562380038387717e-05,
+      "loss": 3.8328,
+      "step": 10260
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.562166773299212e-05,
+      "loss": 3.9218,
+      "step": 10265
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.561953508210706e-05,
+      "loss": 3.9295,
+      "step": 10270
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.561740243122202e-05,
+      "loss": 3.9037,
+      "step": 10275
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.561526978033696e-05,
+      "loss": 4.0196,
+      "step": 10280
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.561313712945191e-05,
+      "loss": 3.9428,
+      "step": 10285
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.561100447856687e-05,
+      "loss": 3.8756,
+      "step": 10290
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.560887182768181e-05,
+      "loss": 3.9362,
+      "step": 10295
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.560673917679676e-05,
+      "loss": 3.931,
+      "step": 10300
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.56046065259117e-05,
+      "loss": 3.9106,
+      "step": 10305
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.560247387502667e-05,
+      "loss": 3.8869,
+      "step": 10310
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.560034122414161e-05,
+      "loss": 3.8791,
+      "step": 10315
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.559820857325657e-05,
+      "loss": 3.9313,
+      "step": 10320
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.559607592237151e-05,
+      "loss": 3.8584,
+      "step": 10325
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.559394327148646e-05,
+      "loss": 3.8393,
+      "step": 10330
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.55918106206014e-05,
+      "loss": 3.9054,
+      "step": 10335
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.558967796971636e-05,
+      "loss": 3.8296,
+      "step": 10340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.55875453188313e-05,
+      "loss": 3.8346,
+      "step": 10345
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.558541266794626e-05,
+      "loss": 3.8221,
+      "step": 10350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.558328001706121e-05,
+      "loss": 3.8941,
+      "step": 10355
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.558114736617617e-05,
+      "loss": 3.8657,
+      "step": 10360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557901471529112e-05,
+      "loss": 3.8482,
+      "step": 10365
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557688206440606e-05,
+      "loss": 3.8533,
+      "step": 10370
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557474941352102e-05,
+      "loss": 3.8012,
+      "step": 10375
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557261676263596e-05,
+      "loss": 3.8421,
+      "step": 10380
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557048411175091e-05,
+      "loss": 3.9025,
+      "step": 10385
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.556835146086585e-05,
+      "loss": 3.8446,
+      "step": 10390
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.556621880998081e-05,
+      "loss": 3.908,
+      "step": 10395
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.556408615909575e-05,
+      "loss": 3.8781,
+      "step": 10400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.556195350821072e-05,
+      "loss": 3.867,
+      "step": 10405
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.555982085732567e-05,
+      "loss": 3.8401,
+      "step": 10410
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.555768820644061e-05,
+      "loss": 3.9424,
+      "step": 10415
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.555555555555557e-05,
+      "loss": 3.8786,
+      "step": 10420
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.555342290467051e-05,
+      "loss": 3.8737,
+      "step": 10425
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.555129025378546e-05,
+      "loss": 3.8781,
+      "step": 10430
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.55491576029004e-05,
+      "loss": 3.7376,
+      "step": 10435
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.554702495201536e-05,
+      "loss": 3.9315,
+      "step": 10440
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.55448923011303e-05,
+      "loss": 3.8474,
+      "step": 10445
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.554275965024526e-05,
+      "loss": 3.8621,
+      "step": 10450
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.554062699936021e-05,
+      "loss": 3.9362,
+      "step": 10455
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.553849434847517e-05,
+      "loss": 3.9791,
+      "step": 10460
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.55363616975901e-05,
+      "loss": 3.8505,
+      "step": 10465
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.553422904670506e-05,
+      "loss": 3.9466,
+      "step": 10470
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.553209639582002e-05,
+      "loss": 3.9119,
+      "step": 10475
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.552996374493496e-05,
+      "loss": 4.0029,
+      "step": 10480
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.552783109404991e-05,
+      "loss": 3.926,
+      "step": 10485
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.552569844316485e-05,
+      "loss": 3.7486,
+      "step": 10490
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.552356579227981e-05,
+      "loss": 3.8645,
+      "step": 10495
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.552143314139475e-05,
+      "loss": 3.8516,
+      "step": 10500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551930049050972e-05,
+      "loss": 3.8728,
+      "step": 10505
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551716783962466e-05,
+      "loss": 3.82,
+      "step": 10510
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551503518873961e-05,
+      "loss": 3.9228,
+      "step": 10515
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551290253785455e-05,
+      "loss": 4.0067,
+      "step": 10520
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551076988696951e-05,
+      "loss": 3.9225,
+      "step": 10525
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.550863723608445e-05,
+      "loss": 3.8593,
+      "step": 10530
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.55065045851994e-05,
+      "loss": 3.8555,
+      "step": 10535
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.550437193431436e-05,
+      "loss": 3.8449,
+      "step": 10540
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.55022392834293e-05,
+      "loss": 3.865,
+      "step": 10545
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.550010663254426e-05,
+      "loss": 3.8232,
+      "step": 10550
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.549797398165921e-05,
+      "loss": 3.8967,
+      "step": 10555
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.549584133077417e-05,
+      "loss": 3.7787,
+      "step": 10560
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.54937086798891e-05,
+      "loss": 3.8804,
+      "step": 10565
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.549157602900406e-05,
+      "loss": 3.8963,
+      "step": 10570
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.5489443378119e-05,
+      "loss": 3.8402,
+      "step": 10575
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.548731072723396e-05,
+      "loss": 3.9114,
+      "step": 10580
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.54851780763489e-05,
+      "loss": 3.9136,
+      "step": 10585
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.548304542546385e-05,
+      "loss": 3.8516,
+      "step": 10590
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.54809127745788e-05,
+      "loss": 3.7779,
+      "step": 10595
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.547878012369376e-05,
+      "loss": 3.8112,
+      "step": 10600
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.547664747280872e-05,
+      "loss": 3.9328,
+      "step": 10605
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.547451482192366e-05,
+      "loss": 3.7798,
+      "step": 10610
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.547238217103861e-05,
+      "loss": 3.9559,
+      "step": 10615
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.547024952015355e-05,
+      "loss": 3.9166,
+      "step": 10620
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.546811686926851e-05,
+      "loss": 3.8668,
+      "step": 10625
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.546598421838345e-05,
+      "loss": 3.856,
+      "step": 10630
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.54638515674984e-05,
+      "loss": 3.9358,
+      "step": 10635
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.546171891661335e-05,
+      "loss": 3.8494,
+      "step": 10640
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.54595862657283e-05,
+      "loss": 3.8878,
+      "step": 10645
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.545745361484326e-05,
+      "loss": 3.8773,
+      "step": 10650
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.545532096395821e-05,
+      "loss": 3.9263,
+      "step": 10655
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.545318831307315e-05,
+      "loss": 3.7892,
+      "step": 10660
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.545105566218811e-05,
+      "loss": 3.7867,
+      "step": 10665
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.544892301130306e-05,
+      "loss": 3.7862,
+      "step": 10670
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.5446790360418e-05,
+      "loss": 3.7521,
+      "step": 10675
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.544465770953296e-05,
+      "loss": 3.9254,
+      "step": 10680
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.54425250586479e-05,
+      "loss": 3.8303,
+      "step": 10685
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.544039240776285e-05,
+      "loss": 3.754,
+      "step": 10690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.543825975687781e-05,
+      "loss": 3.8379,
+      "step": 10695
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.543612710599276e-05,
+      "loss": 3.8989,
+      "step": 10700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.54339944551077e-05,
+      "loss": 3.8993,
+      "step": 10705
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.543186180422266e-05,
+      "loss": 3.8889,
+      "step": 10710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.54297291533376e-05,
+      "loss": 3.869,
+      "step": 10715
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.542759650245255e-05,
+      "loss": 3.8044,
+      "step": 10720
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.54254638515675e-05,
+      "loss": 3.8364,
+      "step": 10725
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.542333120068245e-05,
+      "loss": 3.8113,
+      "step": 10730
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.54211985497974e-05,
+      "loss": 3.8514,
+      "step": 10735
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.541906589891235e-05,
+      "loss": 3.9185,
+      "step": 10740
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.54169332480273e-05,
+      "loss": 3.978,
+      "step": 10745
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.541480059714226e-05,
+      "loss": 3.8926,
+      "step": 10750
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.541266794625721e-05,
+      "loss": 3.8531,
+      "step": 10755
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.541053529537215e-05,
+      "loss": 3.864,
+      "step": 10760
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.540840264448711e-05,
+      "loss": 3.7836,
+      "step": 10765
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.540626999360205e-05,
+      "loss": 3.905,
+      "step": 10770
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.5404137342717e-05,
+      "loss": 3.9142,
+      "step": 10775
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.540200469183194e-05,
+      "loss": 3.832,
+      "step": 10780
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.53998720409469e-05,
+      "loss": 3.8712,
+      "step": 10785
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.539773939006184e-05,
+      "loss": 3.9275,
+      "step": 10790
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.539560673917681e-05,
+      "loss": 3.8709,
+      "step": 10795
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.539347408829176e-05,
+      "loss": 3.8757,
+      "step": 10800
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.53913414374067e-05,
+      "loss": 3.8324,
+      "step": 10805
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.538920878652166e-05,
+      "loss": 3.8187,
+      "step": 10810
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.53870761356366e-05,
+      "loss": 3.8017,
+      "step": 10815
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.538494348475155e-05,
+      "loss": 3.8841,
+      "step": 10820
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.53828108338665e-05,
+      "loss": 3.8557,
+      "step": 10825
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.538067818298145e-05,
+      "loss": 3.8736,
+      "step": 10830
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.537854553209639e-05,
+      "loss": 3.8095,
+      "step": 10835
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.537641288121135e-05,
+      "loss": 3.8223,
+      "step": 10840
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.53742802303263e-05,
+      "loss": 3.7927,
+      "step": 10845
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.537214757944126e-05,
+      "loss": 3.7873,
+      "step": 10850
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.53700149285562e-05,
+      "loss": 3.9492,
+      "step": 10855
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.536788227767115e-05,
+      "loss": 3.8324,
+      "step": 10860
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.536574962678611e-05,
+      "loss": 3.9354,
+      "step": 10865
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.536361697590105e-05,
+      "loss": 3.8118,
+      "step": 10870
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.5361484325016e-05,
+      "loss": 3.8806,
+      "step": 10875
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.535935167413094e-05,
+      "loss": 3.9292,
+      "step": 10880
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.53572190232459e-05,
+      "loss": 3.8313,
+      "step": 10885
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.535508637236085e-05,
+      "loss": 3.9351,
+      "step": 10890
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.535295372147581e-05,
+      "loss": 3.7931,
+      "step": 10895
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.535082107059075e-05,
+      "loss": 3.8909,
+      "step": 10900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.53486884197057e-05,
+      "loss": 3.8385,
+      "step": 10905
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.534655576882065e-05,
+      "loss": 3.8528,
+      "step": 10910
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.53444231179356e-05,
+      "loss": 3.9241,
+      "step": 10915
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.534229046705054e-05,
+      "loss": 3.9389,
+      "step": 10920
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.53401578161655e-05,
+      "loss": 3.9009,
+      "step": 10925
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.533802516528045e-05,
+      "loss": 3.8599,
+      "step": 10930
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.533589251439539e-05,
+      "loss": 3.7557,
+      "step": 10935
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.533375986351035e-05,
+      "loss": 3.8791,
+      "step": 10940
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.53316272126253e-05,
+      "loss": 3.784,
+      "step": 10945
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.532949456174026e-05,
+      "loss": 3.8488,
+      "step": 10950
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.53273619108552e-05,
+      "loss": 3.8906,
+      "step": 10955
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.532522925997015e-05,
+      "loss": 3.9001,
+      "step": 10960
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.53230966090851e-05,
+      "loss": 3.8156,
+      "step": 10965
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.532096395820005e-05,
+      "loss": 3.9544,
+      "step": 10970
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531883130731499e-05,
+      "loss": 3.8833,
+      "step": 10975
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531669865642994e-05,
+      "loss": 3.8381,
+      "step": 10980
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531456600554489e-05,
+      "loss": 3.8852,
+      "step": 10985
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531243335465985e-05,
+      "loss": 3.8482,
+      "step": 10990
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531030070377481e-05,
+      "loss": 3.8442,
+      "step": 10995
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.530816805288975e-05,
+      "loss": 3.9358,
+      "step": 11000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.53060354020047e-05,
+      "loss": 3.8005,
+      "step": 11005
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.530390275111965e-05,
+      "loss": 3.8792,
+      "step": 11010
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.53017701002346e-05,
+      "loss": 3.7995,
+      "step": 11015
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.529963744934954e-05,
+      "loss": 3.8723,
+      "step": 11020
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.52975047984645e-05,
+      "loss": 3.9429,
+      "step": 11025
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.529537214757944e-05,
+      "loss": 3.9988,
+      "step": 11030
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.529323949669439e-05,
+      "loss": 3.9746,
+      "step": 11035
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.529110684580935e-05,
+      "loss": 3.9214,
+      "step": 11040
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.52889741949243e-05,
+      "loss": 3.9161,
+      "step": 11045
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.528684154403924e-05,
+      "loss": 3.8679,
+      "step": 11050
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.52847088931542e-05,
+      "loss": 3.9297,
+      "step": 11055
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.528257624226915e-05,
+      "loss": 3.8636,
+      "step": 11060
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.52804435913841e-05,
+      "loss": 3.8815,
+      "step": 11065
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.527831094049905e-05,
+      "loss": 3.8747,
+      "step": 11070
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.527617828961399e-05,
+      "loss": 3.9002,
+      "step": 11075
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.527404563872894e-05,
+      "loss": 3.8004,
+      "step": 11080
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.52719129878439e-05,
+      "loss": 3.8331,
+      "step": 11085
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.526978033695885e-05,
+      "loss": 3.9091,
+      "step": 11090
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.52676476860738e-05,
+      "loss": 3.8626,
+      "step": 11095
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.526551503518875e-05,
+      "loss": 3.7815,
+      "step": 11100
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.526338238430369e-05,
+      "loss": 3.823,
+      "step": 11105
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.526124973341865e-05,
+      "loss": 3.8166,
+      "step": 11110
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.525911708253359e-05,
+      "loss": 3.8537,
+      "step": 11115
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.525698443164854e-05,
+      "loss": 3.8056,
+      "step": 11120
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.52548517807635e-05,
+      "loss": 3.8709,
+      "step": 11125
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.525271912987844e-05,
+      "loss": 3.8565,
+      "step": 11130
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.525058647899339e-05,
+      "loss": 3.8792,
+      "step": 11135
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.524845382810835e-05,
+      "loss": 3.7237,
+      "step": 11140
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.52463211772233e-05,
+      "loss": 3.8421,
+      "step": 11145
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.524418852633824e-05,
+      "loss": 3.8637,
+      "step": 11150
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.52420558754532e-05,
+      "loss": 3.8359,
+      "step": 11155
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.523992322456814e-05,
+      "loss": 3.8814,
+      "step": 11160
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.52377905736831e-05,
+      "loss": 3.8743,
+      "step": 11165
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.523565792279803e-05,
+      "loss": 3.832,
+      "step": 11170
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.523352527191299e-05,
+      "loss": 3.9288,
+      "step": 11175
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.523139262102793e-05,
+      "loss": 3.8756,
+      "step": 11180
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.52292599701429e-05,
+      "loss": 3.7986,
+      "step": 11185
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.522712731925785e-05,
+      "loss": 3.9345,
+      "step": 11190
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.52249946683728e-05,
+      "loss": 3.8711,
+      "step": 11195
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.522286201748775e-05,
+      "loss": 3.7496,
+      "step": 11200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.522072936660269e-05,
+      "loss": 3.8577,
+      "step": 11205
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.521859671571765e-05,
+      "loss": 3.8267,
+      "step": 11210
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.521646406483259e-05,
+      "loss": 3.8628,
+      "step": 11215
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.521433141394754e-05,
+      "loss": 3.7818,
+      "step": 11220
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.521219876306248e-05,
+      "loss": 3.8626,
+      "step": 11225
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.521006611217744e-05,
+      "loss": 3.7875,
+      "step": 11230
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.520793346129239e-05,
+      "loss": 3.8996,
+      "step": 11235
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.520580081040735e-05,
+      "loss": 3.922,
+      "step": 11240
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.520366815952229e-05,
+      "loss": 3.9729,
+      "step": 11245
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.520153550863724e-05,
+      "loss": 3.8814,
+      "step": 11250
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.51994028577522e-05,
+      "loss": 3.8322,
+      "step": 11255
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.519727020686714e-05,
+      "loss": 3.9059,
+      "step": 11260
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.51951375559821e-05,
+      "loss": 3.8843,
+      "step": 11265
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.519300490509703e-05,
+      "loss": 3.8554,
+      "step": 11270
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.519087225421199e-05,
+      "loss": 3.917,
+      "step": 11275
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.518873960332694e-05,
+      "loss": 3.7869,
+      "step": 11280
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.51866069524419e-05,
+      "loss": 3.7851,
+      "step": 11285
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.518447430155684e-05,
+      "loss": 3.8404,
+      "step": 11290
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.51823416506718e-05,
+      "loss": 3.8633,
+      "step": 11295
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.518020899978674e-05,
+      "loss": 3.8489,
+      "step": 11300
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.517807634890169e-05,
+      "loss": 3.8636,
+      "step": 11305
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.517594369801663e-05,
+      "loss": 3.9264,
+      "step": 11310
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.517381104713159e-05,
+      "loss": 3.868,
+      "step": 11315
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.517167839624654e-05,
+      "loss": 3.9043,
+      "step": 11320
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.516954574536148e-05,
+      "loss": 3.8982,
+      "step": 11325
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.516741309447644e-05,
+      "loss": 3.8716,
+      "step": 11330
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.516528044359139e-05,
+      "loss": 3.8573,
+      "step": 11335
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.516314779270635e-05,
+      "loss": 3.8715,
+      "step": 11340
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.516101514182129e-05,
+      "loss": 3.9904,
+      "step": 11345
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.515888249093624e-05,
+      "loss": 3.8789,
+      "step": 11350
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.515674984005118e-05,
+      "loss": 3.8145,
+      "step": 11355
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.515461718916614e-05,
+      "loss": 3.8247,
+      "step": 11360
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.515248453828108e-05,
+      "loss": 3.9156,
+      "step": 11365
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.515035188739603e-05,
+      "loss": 3.8972,
+      "step": 11370
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.514821923651099e-05,
+      "loss": 3.8997,
+      "step": 11375
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.514608658562594e-05,
+      "loss": 3.9253,
+      "step": 11380
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.51439539347409e-05,
+      "loss": 3.8999,
+      "step": 11385
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.514182128385584e-05,
+      "loss": 3.9728,
+      "step": 11390
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.51396886329708e-05,
+      "loss": 3.8845,
+      "step": 11395
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.513755598208574e-05,
+      "loss": 3.854,
+      "step": 11400
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.513542333120069e-05,
+      "loss": 3.8744,
+      "step": 11405
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.513329068031563e-05,
+      "loss": 3.8516,
+      "step": 11410
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.513115802943059e-05,
+      "loss": 3.7825,
+      "step": 11415
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.512902537854553e-05,
+      "loss": 3.8688,
+      "step": 11420
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.512689272766048e-05,
+      "loss": 3.8947,
+      "step": 11425
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.512476007677544e-05,
+      "loss": 3.8144,
+      "step": 11430
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.512262742589039e-05,
+      "loss": 3.9057,
+      "step": 11435
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.512049477500533e-05,
+      "loss": 3.9402,
+      "step": 11440
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.511836212412029e-05,
+      "loss": 3.8584,
+      "step": 11445
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.511622947323524e-05,
+      "loss": 3.9328,
+      "step": 11450
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.511409682235018e-05,
+      "loss": 3.9059,
+      "step": 11455
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.511196417146514e-05,
+      "loss": 3.887,
+      "step": 11460
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.510983152058008e-05,
+      "loss": 3.8703,
+      "step": 11465
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.510769886969503e-05,
+      "loss": 3.9565,
+      "step": 11470
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.510556621880999e-05,
+      "loss": 3.9683,
+      "step": 11475
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.510343356792494e-05,
+      "loss": 3.8323,
+      "step": 11480
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.510130091703989e-05,
+      "loss": 3.9019,
+      "step": 11485
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.509916826615484e-05,
+      "loss": 3.9326,
+      "step": 11490
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.509703561526978e-05,
+      "loss": 3.8418,
+      "step": 11495
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.509490296438474e-05,
+      "loss": 3.941,
+      "step": 11500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.509277031349968e-05,
+      "loss": 3.8944,
+      "step": 11505
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.509063766261463e-05,
+      "loss": 3.8895,
+      "step": 11510
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.508850501172959e-05,
+      "loss": 3.877,
+      "step": 11515
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.508637236084453e-05,
+      "loss": 3.8945,
+      "step": 11520
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.508423970995948e-05,
+      "loss": 3.9132,
+      "step": 11525
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.508210705907444e-05,
+      "loss": 3.8479,
+      "step": 11530
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.507997440818939e-05,
+      "loss": 3.9395,
+      "step": 11535
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.507784175730433e-05,
+      "loss": 3.7716,
+      "step": 11540
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.507570910641929e-05,
+      "loss": 4.0152,
+      "step": 11545
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.507357645553423e-05,
+      "loss": 3.8792,
+      "step": 11550
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.507144380464918e-05,
+      "loss": 3.8406,
+      "step": 11555
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.506931115376413e-05,
+      "loss": 3.8779,
+      "step": 11560
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.506717850287908e-05,
+      "loss": 3.8604,
+      "step": 11565
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.506504585199403e-05,
+      "loss": 3.9058,
+      "step": 11570
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.506291320110899e-05,
+      "loss": 3.8959,
+      "step": 11575
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.506078055022394e-05,
+      "loss": 3.9086,
+      "step": 11580
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.505864789933889e-05,
+      "loss": 3.8602,
+      "step": 11585
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.505651524845384e-05,
+      "loss": 3.9418,
+      "step": 11590
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.505438259756878e-05,
+      "loss": 3.8621,
+      "step": 11595
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.505224994668374e-05,
+      "loss": 3.891,
+      "step": 11600
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.505011729579868e-05,
+      "loss": 3.8971,
+      "step": 11605
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.504798464491363e-05,
+      "loss": 3.8623,
+      "step": 11610
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.504585199402857e-05,
+      "loss": 3.8798,
+      "step": 11615
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.504371934314353e-05,
+      "loss": 3.8488,
+      "step": 11620
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.504158669225848e-05,
+      "loss": 3.7583,
+      "step": 11625
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.503945404137344e-05,
+      "loss": 3.833,
+      "step": 11630
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.503732139048838e-05,
+      "loss": 3.8027,
+      "step": 11635
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.503518873960333e-05,
+      "loss": 3.7856,
+      "step": 11640
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.503305608871829e-05,
+      "loss": 3.818,
+      "step": 11645
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.503092343783323e-05,
+      "loss": 3.9409,
+      "step": 11650
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.502879078694818e-05,
+      "loss": 3.8591,
+      "step": 11655
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.502665813606313e-05,
+      "loss": 3.861,
+      "step": 11660
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.502452548517808e-05,
+      "loss": 4.0183,
+      "step": 11665
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.502239283429303e-05,
+      "loss": 3.9115,
+      "step": 11670
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.502026018340799e-05,
+      "loss": 3.8273,
+      "step": 11675
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.501812753252293e-05,
+      "loss": 3.8538,
+      "step": 11680
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.501599488163789e-05,
+      "loss": 3.7904,
+      "step": 11685
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.501386223075283e-05,
+      "loss": 3.8011,
+      "step": 11690
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.501172957986778e-05,
+      "loss": 3.7745,
+      "step": 11695
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.500959692898272e-05,
+      "loss": 3.8577,
+      "step": 11700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.500746427809768e-05,
+      "loss": 3.8567,
+      "step": 11705
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.500533162721263e-05,
+      "loss": 3.8914,
+      "step": 11710
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.500319897632757e-05,
+      "loss": 3.8989,
+      "step": 11715
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.500106632544253e-05,
+      "loss": 3.815,
+      "step": 11720
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499893367455748e-05,
+      "loss": 3.8976,
+      "step": 11725
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499680102367244e-05,
+      "loss": 3.9922,
+      "step": 11730
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499466837278738e-05,
+      "loss": 3.8147,
+      "step": 11735
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499253572190233e-05,
+      "loss": 3.8722,
+      "step": 11740
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499040307101727e-05,
+      "loss": 3.9046,
+      "step": 11745
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.498827042013223e-05,
+      "loss": 3.9106,
+      "step": 11750
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.498613776924717e-05,
+      "loss": 3.9781,
+      "step": 11755
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.498400511836213e-05,
+      "loss": 3.7725,
+      "step": 11760
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.498187246747708e-05,
+      "loss": 4.0173,
+      "step": 11765
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.497973981659203e-05,
+      "loss": 3.8014,
+      "step": 11770
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.497760716570699e-05,
+      "loss": 3.8209,
+      "step": 11775
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.497547451482193e-05,
+      "loss": 3.7884,
+      "step": 11780
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.497334186393689e-05,
+      "loss": 3.9114,
+      "step": 11785
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.497120921305183e-05,
+      "loss": 3.8205,
+      "step": 11790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496907656216678e-05,
+      "loss": 3.8869,
+      "step": 11795
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496694391128172e-05,
+      "loss": 3.847,
+      "step": 11800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496481126039668e-05,
+      "loss": 3.9064,
+      "step": 11805
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496267860951162e-05,
+      "loss": 3.8412,
+      "step": 11810
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496054595862657e-05,
+      "loss": 3.882,
+      "step": 11815
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.495841330774153e-05,
+      "loss": 3.9267,
+      "step": 11820
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.495628065685648e-05,
+      "loss": 3.9251,
+      "step": 11825
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.495414800597142e-05,
+      "loss": 3.8157,
+      "step": 11830
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.495201535508638e-05,
+      "loss": 3.8826,
+      "step": 11835
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.494988270420133e-05,
+      "loss": 3.7828,
+      "step": 11840
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.494775005331627e-05,
+      "loss": 3.8646,
+      "step": 11845
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.494561740243123e-05,
+      "loss": 3.9115,
+      "step": 11850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.494348475154617e-05,
+      "loss": 3.8406,
+      "step": 11855
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.494135210066113e-05,
+      "loss": 3.8486,
+      "step": 11860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.493921944977608e-05,
+      "loss": 3.7279,
+      "step": 11865
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.493708679889103e-05,
+      "loss": 3.8669,
+      "step": 11870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.493495414800598e-05,
+      "loss": 3.762,
+      "step": 11875
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.493282149712093e-05,
+      "loss": 3.967,
+      "step": 11880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.493068884623587e-05,
+      "loss": 3.8016,
+      "step": 11885
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.492855619535083e-05,
+      "loss": 3.7951,
+      "step": 11890
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.492642354446577e-05,
+      "loss": 3.9575,
+      "step": 11895
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.492429089358072e-05,
+      "loss": 3.8486,
+      "step": 11900
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.492215824269568e-05,
+      "loss": 3.8549,
+      "step": 11905
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.492002559181062e-05,
+      "loss": 3.7932,
+      "step": 11910
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.491789294092557e-05,
+      "loss": 3.7255,
+      "step": 11915
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.491576029004053e-05,
+      "loss": 3.8825,
+      "step": 11920
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.491362763915548e-05,
+      "loss": 3.8665,
+      "step": 11925
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.491149498827042e-05,
+      "loss": 3.9885,
+      "step": 11930
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.490936233738538e-05,
+      "loss": 3.9467,
+      "step": 11935
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.490722968650032e-05,
+      "loss": 3.9258,
+      "step": 11940
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.490509703561527e-05,
+      "loss": 3.7383,
+      "step": 11945
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.490296438473022e-05,
+      "loss": 3.8705,
+      "step": 11950
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.490083173384517e-05,
+      "loss": 3.773,
+      "step": 11955
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.489869908296013e-05,
+      "loss": 3.838,
+      "step": 11960
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.489656643207508e-05,
+      "loss": 3.801,
+      "step": 11965
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.489443378119003e-05,
+      "loss": 3.7859,
+      "step": 11970
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.489230113030498e-05,
+      "loss": 3.9198,
+      "step": 11975
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.489016847941993e-05,
+      "loss": 3.8493,
+      "step": 11980
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.488803582853487e-05,
+      "loss": 3.888,
+      "step": 11985
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.488590317764983e-05,
+      "loss": 3.8691,
+      "step": 11990
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.488377052676477e-05,
+      "loss": 3.7743,
+      "step": 11995
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.488163787587972e-05,
+      "loss": 3.9196,
+      "step": 12000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.487950522499466e-05,
+      "loss": 3.8054,
+      "step": 12005
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.487737257410962e-05,
+      "loss": 3.8614,
+      "step": 12010
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.487523992322457e-05,
+      "loss": 3.8772,
+      "step": 12015
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.487310727233953e-05,
+      "loss": 3.7921,
+      "step": 12020
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.487097462145447e-05,
+      "loss": 3.8524,
+      "step": 12025
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.486884197056942e-05,
+      "loss": 3.8269,
+      "step": 12030
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.486670931968438e-05,
+      "loss": 3.877,
+      "step": 12035
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.486457666879932e-05,
+      "loss": 3.8248,
+      "step": 12040
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.486244401791427e-05,
+      "loss": 3.8146,
+      "step": 12045
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.486031136702922e-05,
+      "loss": 3.8214,
+      "step": 12050
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.485817871614417e-05,
+      "loss": 3.8447,
+      "step": 12055
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.485604606525913e-05,
+      "loss": 3.8957,
+      "step": 12060
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.485391341437408e-05,
+      "loss": 3.81,
+      "step": 12065
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.485178076348902e-05,
+      "loss": 3.8389,
+      "step": 12070
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.484964811260398e-05,
+      "loss": 3.8421,
+      "step": 12075
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.484751546171892e-05,
+      "loss": 3.8608,
+      "step": 12080
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.484538281083387e-05,
+      "loss": 3.8778,
+      "step": 12085
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.484325015994881e-05,
+      "loss": 3.8864,
+      "step": 12090
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.484111750906377e-05,
+      "loss": 3.9188,
+      "step": 12095
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.483898485817872e-05,
+      "loss": 3.9404,
+      "step": 12100
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.483685220729366e-05,
+      "loss": 3.9068,
+      "step": 12105
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.483471955640862e-05,
+      "loss": 3.8319,
+      "step": 12110
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.483258690552357e-05,
+      "loss": 3.8308,
+      "step": 12115
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.483045425463853e-05,
+      "loss": 3.8978,
+      "step": 12120
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.482832160375347e-05,
+      "loss": 3.8876,
+      "step": 12125
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.482618895286842e-05,
+      "loss": 3.8715,
+      "step": 12130
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.482405630198337e-05,
+      "loss": 3.8699,
+      "step": 12135
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.482192365109832e-05,
+      "loss": 3.7837,
+      "step": 12140
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481979100021326e-05,
+      "loss": 3.8165,
+      "step": 12145
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481765834932822e-05,
+      "loss": 3.8433,
+      "step": 12150
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481552569844317e-05,
+      "loss": 3.8599,
+      "step": 12155
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481339304755813e-05,
+      "loss": 3.8229,
+      "step": 12160
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481126039667308e-05,
+      "loss": 3.929,
+      "step": 12165
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.480912774578802e-05,
+      "loss": 3.869,
+      "step": 12170
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.480699509490298e-05,
+      "loss": 3.8685,
+      "step": 12175
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.480486244401792e-05,
+      "loss": 3.9377,
+      "step": 12180
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.480272979313287e-05,
+      "loss": 3.8776,
+      "step": 12185
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.480059714224781e-05,
+      "loss": 3.8821,
+      "step": 12190
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.479846449136277e-05,
+      "loss": 3.8149,
+      "step": 12195
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.479633184047771e-05,
+      "loss": 3.8796,
+      "step": 12200
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.479419918959266e-05,
+      "loss": 3.8825,
+      "step": 12205
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.479206653870762e-05,
+      "loss": 3.8366,
+      "step": 12210
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.478993388782257e-05,
+      "loss": 3.8682,
+      "step": 12215
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.478780123693751e-05,
+      "loss": 4.0087,
+      "step": 12220
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.478566858605247e-05,
+      "loss": 3.8197,
+      "step": 12225
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.478353593516742e-05,
+      "loss": 3.96,
+      "step": 12230
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.478140328428237e-05,
+      "loss": 3.7963,
+      "step": 12235
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.477927063339732e-05,
+      "loss": 3.8428,
+      "step": 12240
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.477713798251226e-05,
+      "loss": 3.8662,
+      "step": 12245
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.477500533162722e-05,
+      "loss": 3.8656,
+      "step": 12250
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.477287268074217e-05,
+      "loss": 3.8357,
+      "step": 12255
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.477074002985713e-05,
+      "loss": 3.8721,
+      "step": 12260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.476860737897207e-05,
+      "loss": 3.8898,
+      "step": 12265
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.476647472808702e-05,
+      "loss": 3.853,
+      "step": 12270
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.476434207720196e-05,
+      "loss": 3.8396,
+      "step": 12275
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.476220942631692e-05,
+      "loss": 3.7641,
+      "step": 12280
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.476007677543186e-05,
+      "loss": 3.7911,
+      "step": 12285
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.475794412454681e-05,
+      "loss": 3.8446,
+      "step": 12290
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.475581147366177e-05,
+      "loss": 3.8506,
+      "step": 12295
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.475367882277671e-05,
+      "loss": 3.87,
+      "step": 12300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.475154617189166e-05,
+      "loss": 3.7995,
+      "step": 12305
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.474941352100662e-05,
+      "loss": 3.8378,
+      "step": 12310
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.474728087012157e-05,
+      "loss": 3.8838,
+      "step": 12315
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.474514821923651e-05,
+      "loss": 3.8849,
+      "step": 12320
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.474301556835147e-05,
+      "loss": 3.8572,
+      "step": 12325
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.474088291746641e-05,
+      "loss": 3.7967,
+      "step": 12330
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.473875026658137e-05,
+      "loss": 3.9313,
+      "step": 12335
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.47366176156963e-05,
+      "loss": 3.9456,
+      "step": 12340
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.473448496481126e-05,
+      "loss": 3.7697,
+      "step": 12345
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.473235231392622e-05,
+      "loss": 3.9275,
+      "step": 12350
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.473021966304117e-05,
+      "loss": 3.8399,
+      "step": 12355
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.472808701215613e-05,
+      "loss": 3.8394,
+      "step": 12360
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.472595436127107e-05,
+      "loss": 3.8439,
+      "step": 12365
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.472382171038602e-05,
+      "loss": 3.9485,
+      "step": 12370
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.472168905950096e-05,
+      "loss": 3.9012,
+      "step": 12375
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.471955640861592e-05,
+      "loss": 3.8532,
+      "step": 12380
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.471742375773086e-05,
+      "loss": 3.8111,
+      "step": 12385
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.471529110684581e-05,
+      "loss": 3.8193,
+      "step": 12390
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.471315845596075e-05,
+      "loss": 3.7686,
+      "step": 12395
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.471102580507571e-05,
+      "loss": 3.8529,
+      "step": 12400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470889315419066e-05,
+      "loss": 3.9288,
+      "step": 12405
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470676050330562e-05,
+      "loss": 3.8933,
+      "step": 12410
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470462785242057e-05,
+      "loss": 3.9774,
+      "step": 12415
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470249520153551e-05,
+      "loss": 3.8566,
+      "step": 12420
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470036255065047e-05,
+      "loss": 3.8747,
+      "step": 12425
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.469822989976541e-05,
+      "loss": 3.8786,
+      "step": 12430
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.469609724888037e-05,
+      "loss": 3.8393,
+      "step": 12435
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.46939645979953e-05,
+      "loss": 3.7922,
+      "step": 12440
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.469183194711026e-05,
+      "loss": 3.7708,
+      "step": 12445
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.468969929622522e-05,
+      "loss": 3.9003,
+      "step": 12450
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468756664534017e-05,
+      "loss": 3.8484,
+      "step": 12455
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468543399445511e-05,
+      "loss": 3.8923,
+      "step": 12460
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468330134357007e-05,
+      "loss": 3.8726,
+      "step": 12465
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468116869268501e-05,
+      "loss": 3.8075,
+      "step": 12470
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467903604179996e-05,
+      "loss": 3.8379,
+      "step": 12475
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467690339091492e-05,
+      "loss": 3.9067,
+      "step": 12480
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467477074002986e-05,
+      "loss": 3.8481,
+      "step": 12485
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467263808914481e-05,
+      "loss": 3.8343,
+      "step": 12490
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467050543825975e-05,
+      "loss": 3.8555,
+      "step": 12495
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.466837278737471e-05,
+      "loss": 3.9263,
+      "step": 12500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.466624013648966e-05,
+      "loss": 3.8103,
+      "step": 12505
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.466410748560462e-05,
+      "loss": 3.8817,
+      "step": 12510
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.466197483471956e-05,
+      "loss": 3.7999,
+      "step": 12515
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.465984218383451e-05,
+      "loss": 3.8179,
+      "step": 12520
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.465770953294946e-05,
+      "loss": 3.8504,
+      "step": 12525
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.465557688206441e-05,
+      "loss": 3.8828,
+      "step": 12530
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.465344423117935e-05,
+      "loss": 3.8332,
+      "step": 12535
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.46513115802943e-05,
+      "loss": 3.9158,
+      "step": 12540
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.464917892940926e-05,
+      "loss": 3.8512,
+      "step": 12545
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.464704627852422e-05,
+      "loss": 3.8978,
+      "step": 12550
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.464491362763917e-05,
+      "loss": 3.8572,
+      "step": 12555
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.464278097675411e-05,
+      "loss": 3.933,
+      "step": 12560
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.464064832586907e-05,
+      "loss": 3.8698,
+      "step": 12565
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.463851567498401e-05,
+      "loss": 3.7444,
+      "step": 12570
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.463638302409896e-05,
+      "loss": 3.8093,
+      "step": 12575
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.46342503732139e-05,
+      "loss": 3.9234,
+      "step": 12580
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.463211772232886e-05,
+      "loss": 3.8823,
+      "step": 12585
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.46299850714438e-05,
+      "loss": 3.9563,
+      "step": 12590
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.462785242055875e-05,
+      "loss": 3.8849,
+      "step": 12595
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.462571976967371e-05,
+      "loss": 3.8896,
+      "step": 12600
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.462358711878866e-05,
+      "loss": 3.8194,
+      "step": 12605
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.462145446790362e-05,
+      "loss": 3.8281,
+      "step": 12610
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.461932181701856e-05,
+      "loss": 3.7781,
+      "step": 12615
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.461718916613351e-05,
+      "loss": 3.8643,
+      "step": 12620
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.461505651524846e-05,
+      "loss": 3.8265,
+      "step": 12625
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.461292386436341e-05,
+      "loss": 3.8002,
+      "step": 12630
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.461079121347835e-05,
+      "loss": 3.8869,
+      "step": 12635
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.46086585625933e-05,
+      "loss": 3.8442,
+      "step": 12640
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.460652591170826e-05,
+      "loss": 3.8418,
+      "step": 12645
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.460439326082322e-05,
+      "loss": 3.7834,
+      "step": 12650
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.460226060993816e-05,
+      "loss": 3.8669,
+      "step": 12655
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.460012795905311e-05,
+      "loss": 3.7716,
+      "step": 12660
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.459799530816805e-05,
+      "loss": 3.8248,
+      "step": 12665
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.459586265728301e-05,
+      "loss": 3.8009,
+      "step": 12670
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.459373000639796e-05,
+      "loss": 3.884,
+      "step": 12675
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.45915973555129e-05,
+      "loss": 3.7295,
+      "step": 12680
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.458946470462786e-05,
+      "loss": 3.9192,
+      "step": 12685
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.45873320537428e-05,
+      "loss": 3.8366,
+      "step": 12690
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.458519940285775e-05,
+      "loss": 3.8342,
+      "step": 12695
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.458306675197271e-05,
+      "loss": 3.8637,
+      "step": 12700
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.458093410108766e-05,
+      "loss": 3.7581,
+      "step": 12705
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.45788014502026e-05,
+      "loss": 3.8351,
+      "step": 12710
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.457666879931756e-05,
+      "loss": 3.8001,
+      "step": 12715
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.45745361484325e-05,
+      "loss": 3.7676,
+      "step": 12720
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.457240349754746e-05,
+      "loss": 3.8946,
+      "step": 12725
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.45702708466624e-05,
+      "loss": 3.9437,
+      "step": 12730
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.456813819577735e-05,
+      "loss": 3.7802,
+      "step": 12735
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.45660055448923e-05,
+      "loss": 3.8232,
+      "step": 12740
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.456387289400726e-05,
+      "loss": 3.7918,
+      "step": 12745
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.456174024312222e-05,
+      "loss": 3.7892,
+      "step": 12750
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.455960759223716e-05,
+      "loss": 3.9601,
+      "step": 12755
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.455747494135211e-05,
+      "loss": 3.9499,
+      "step": 12760
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.455534229046705e-05,
+      "loss": 3.886,
+      "step": 12765
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.455320963958201e-05,
+      "loss": 3.8193,
+      "step": 12770
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.455107698869695e-05,
+      "loss": 3.8418,
+      "step": 12775
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.45489443378119e-05,
+      "loss": 3.7722,
+      "step": 12780
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.454681168692684e-05,
+      "loss": 3.8784,
+      "step": 12785
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.45446790360418e-05,
+      "loss": 3.8373,
+      "step": 12790
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.454254638515675e-05,
+      "loss": 3.7727,
+      "step": 12795
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.454041373427171e-05,
+      "loss": 3.8564,
+      "step": 12800
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.453828108338666e-05,
+      "loss": 3.7502,
+      "step": 12805
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.45361484325016e-05,
+      "loss": 3.8341,
+      "step": 12810
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.453401578161656e-05,
+      "loss": 3.757,
+      "step": 12815
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.45318831307315e-05,
+      "loss": 3.7438,
+      "step": 12820
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.452975047984646e-05,
+      "loss": 3.7876,
+      "step": 12825
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.45276178289614e-05,
+      "loss": 3.7533,
+      "step": 12830
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.452548517807635e-05,
+      "loss": 3.7903,
+      "step": 12835
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.45233525271913e-05,
+      "loss": 3.8345,
+      "step": 12840
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.452121987630626e-05,
+      "loss": 3.8461,
+      "step": 12845
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.45190872254212e-05,
+      "loss": 3.8457,
+      "step": 12850
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.451695457453616e-05,
+      "loss": 3.8823,
+      "step": 12855
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.45148219236511e-05,
+      "loss": 3.8906,
+      "step": 12860
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.451268927276605e-05,
+      "loss": 3.8835,
+      "step": 12865
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.451055662188101e-05,
+      "loss": 3.8094,
+      "step": 12870
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.450842397099595e-05,
+      "loss": 3.9155,
+      "step": 12875
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.45062913201109e-05,
+      "loss": 3.8129,
+      "step": 12880
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.450415866922584e-05,
+      "loss": 3.7896,
+      "step": 12885
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.45020260183408e-05,
+      "loss": 3.9116,
+      "step": 12890
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.449989336745575e-05,
+      "loss": 3.785,
+      "step": 12895
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.449776071657071e-05,
+      "loss": 3.8062,
+      "step": 12900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.449562806568565e-05,
+      "loss": 3.8623,
+      "step": 12905
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.44934954148006e-05,
+      "loss": 3.9554,
+      "step": 12910
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.449136276391555e-05,
+      "loss": 3.8695,
+      "step": 12915
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.44892301130305e-05,
+      "loss": 3.872,
+      "step": 12920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.448709746214544e-05,
+      "loss": 3.7832,
+      "step": 12925
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.44849648112604e-05,
+      "loss": 3.9063,
+      "step": 12930
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.448283216037535e-05,
+      "loss": 3.8449,
+      "step": 12935
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.44806995094903e-05,
+      "loss": 3.8684,
+      "step": 12940
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.447856685860526e-05,
+      "loss": 3.8864,
+      "step": 12945
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.44764342077202e-05,
+      "loss": 3.8493,
+      "step": 12950
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.447430155683516e-05,
+      "loss": 3.7985,
+      "step": 12955
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.44721689059501e-05,
+      "loss": 3.9547,
+      "step": 12960
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.447003625506505e-05,
+      "loss": 3.8626,
+      "step": 12965
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.446790360418e-05,
+      "loss": 3.7944,
+      "step": 12970
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.446577095329495e-05,
+      "loss": 3.8526,
+      "step": 12975
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.446363830240989e-05,
+      "loss": 3.8157,
+      "step": 12980
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.446150565152484e-05,
+      "loss": 3.8383,
+      "step": 12985
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.44593730006398e-05,
+      "loss": 3.8873,
+      "step": 12990
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.445724034975475e-05,
+      "loss": 3.9132,
+      "step": 12995
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.445510769886971e-05,
+      "loss": 3.7766,
+      "step": 13000
     }
   ],
   "logging_steps": 5,
-  "max_steps": 1945,
-  "num_train_epochs": 5,
+  "max_steps": 234450,
+  "num_train_epochs": 50,
   "save_steps": 1000,
-  "total_flos": 1824768000000000.0,
+  "total_flos": 4.56192e+16,
   "trial_name": null,
   "trial_params": null
 }