diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -10,1639 +10,1639 @@
     {
       "epoch": 0.0,
       "learning_rate": 1.9998144540309864e-05,
-      "loss": 0.6623,
+      "loss": 0.6522,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9996289080619725e-05,
-      "loss": 0.5762,
+      "loss": 0.5273,
       "step": 2
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9994433620929587e-05,
-      "loss": 0.483,
+      "loss": 0.4584,
       "step": 3
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.999257816123945e-05,
-      "loss": 0.3882,
+      "loss": 0.3972,
       "step": 4
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.999072270154931e-05,
-      "loss": 0.3943,
+      "loss": 0.295,
       "step": 5
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.998886724185917e-05,
-      "loss": 0.2874,
+      "loss": 0.2801,
       "step": 6
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9987011782169035e-05,
-      "loss": 0.259,
+      "loss": 0.2672,
       "step": 7
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9985156322478897e-05,
-      "loss": 0.2403,
+      "loss": 0.2612,
       "step": 8
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.998330086278876e-05,
-      "loss": 0.1743,
+      "loss": 0.1866,
       "step": 9
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9981445403098617e-05,
-      "loss": 0.2289,
+      "loss": 0.1299,
       "step": 10
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9979589943408482e-05,
-      "loss": 0.1836,
+      "loss": 0.1314,
       "step": 11
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9977734483718344e-05,
-      "loss": 0.2114,
+      "loss": 0.1452,
       "step": 12
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9975879024028203e-05,
-      "loss": 0.133,
+      "loss": 0.1988,
       "step": 13
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.9974023564338064e-05,
-      "loss": 0.1171,
+      "loss": 0.2823,
       "step": 14
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.997216810464793e-05,
-      "loss": 0.0704,
+      "loss": 0.2078,
       "step": 15
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.997031264495779e-05,
-      "loss": 0.1956,
+      "loss": 0.069,
       "step": 16
     },
     {
       "epoch": 0.0,
       "learning_rate": 1.996845718526765e-05,
-      "loss": 0.1071,
+      "loss": 0.1251,
       "step": 17
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9966601725577512e-05,
-      "loss": 0.4079,
+      "loss": 0.2065,
       "step": 18
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9964746265887377e-05,
-      "loss": 0.1876,
+      "loss": 0.1458,
       "step": 19
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9962890806197236e-05,
-      "loss": 0.1395,
+      "loss": 0.1095,
       "step": 20
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9961035346507098e-05,
-      "loss": 0.049,
+      "loss": 0.0496,
       "step": 21
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.995917988681696e-05,
-      "loss": 0.1874,
+      "loss": 0.0455,
       "step": 22
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.995732442712682e-05,
-      "loss": 0.3381,
+      "loss": 0.1868,
       "step": 23
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9955468967436683e-05,
-      "loss": 0.0913,
+      "loss": 0.1877,
       "step": 24
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9953613507746545e-05,
-      "loss": 0.1867,
+      "loss": 0.0915,
       "step": 25
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9951758048056407e-05,
-      "loss": 0.29,
+      "loss": 0.0918,
       "step": 26
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.994990258836627e-05,
-      "loss": 0.0973,
+      "loss": 0.1434,
       "step": 27
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.994804712867613e-05,
-      "loss": 0.1436,
+      "loss": 0.1397,
       "step": 28
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9946191668985992e-05,
-      "loss": 0.286,
+      "loss": 0.2614,
       "step": 29
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9944336209295854e-05,
-      "loss": 0.1422,
+      "loss": 0.16,
       "step": 30
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9942480749605716e-05,
-      "loss": 0.2818,
+      "loss": 0.1976,
       "step": 31
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9940625289915578e-05,
-      "loss": 0.2774,
+      "loss": 0.0534,
       "step": 32
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.993876983022544e-05,
-      "loss": 0.2312,
+      "loss": 0.0352,
       "step": 33
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9936914370535302e-05,
-      "loss": 0.234,
+      "loss": 0.2084,
       "step": 34
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9935058910845164e-05,
-      "loss": 0.1564,
+      "loss": 0.4373,
       "step": 35
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9933203451155026e-05,
-      "loss": 0.1598,
+      "loss": 0.0411,
       "step": 36
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9931347991464887e-05,
-      "loss": 0.0659,
+      "loss": 0.2488,
       "step": 37
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.992949253177475e-05,
-      "loss": 0.0995,
+      "loss": 0.1537,
       "step": 38
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.992763707208461e-05,
-      "loss": 0.2888,
+      "loss": 0.0384,
       "step": 39
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9925781612394473e-05,
-      "loss": 0.0935,
+      "loss": 0.0461,
       "step": 40
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9923926152704335e-05,
-      "loss": 0.1428,
+      "loss": 0.0319,
       "step": 41
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9922070693014193e-05,
-      "loss": 0.0899,
+      "loss": 0.1531,
       "step": 42
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.992021523332406e-05,
-      "loss": 0.1406,
+      "loss": 0.2518,
       "step": 43
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.991835977363392e-05,
-      "loss": 0.1905,
+      "loss": 0.0546,
       "step": 44
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9916504313943782e-05,
-      "loss": 0.1399,
+      "loss": 0.1454,
       "step": 45
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.991464885425364e-05,
-      "loss": 0.2441,
+      "loss": 0.0933,
       "step": 46
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9912793394563506e-05,
-      "loss": 0.2373,
+      "loss": 0.207,
       "step": 47
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9910937934873368e-05,
-      "loss": 0.091,
+      "loss": 0.309,
       "step": 48
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9909082475183227e-05,
-      "loss": 0.1377,
+      "loss": 0.0894,
       "step": 49
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.990722701549309e-05,
-      "loss": 0.1371,
+      "loss": 0.1448,
       "step": 50
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9905371555802954e-05,
-      "loss": 0.0897,
+      "loss": 0.1871,
       "step": 51
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9903516096112812e-05,
-      "loss": 0.239,
+      "loss": 0.0539,
       "step": 52
     },
     {
       "epoch": 0.01,
       "learning_rate": 1.9901660636422674e-05,
-      "loss": 0.1415,
+      "loss": 0.0892,
       "step": 53
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9899805176732536e-05,
-      "loss": 0.1898,
+      "loss": 0.1967,
       "step": 54
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.98979497170424e-05,
-      "loss": 0.1907,
+      "loss": 0.0312,
       "step": 55
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.989609425735226e-05,
-      "loss": 0.142,
+      "loss": 0.2541,
       "step": 56
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.989423879766212e-05,
-      "loss": 0.1422,
+      "loss": 0.1481,
       "step": 57
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9892383337971983e-05,
-      "loss": 0.1345,
+      "loss": 0.2065,
       "step": 58
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9890527878281845e-05,
-      "loss": 0.2453,
+      "loss": 0.1939,
       "step": 59
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9888672418591707e-05,
-      "loss": 0.0894,
+      "loss": 0.032,
       "step": 60
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.988681695890157e-05,
-      "loss": 0.0883,
+      "loss": 0.0341,
       "step": 61
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.988496149921143e-05,
-      "loss": 0.1406,
+      "loss": 0.0836,
       "step": 62
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9883106039521293e-05,
-      "loss": 0.1939,
+      "loss": 0.1426,
       "step": 63
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9881250579831155e-05,
-      "loss": 0.1909,
+      "loss": 0.1994,
       "step": 64
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9879395120141016e-05,
-      "loss": 0.2411,
+      "loss": 0.0891,
       "step": 65
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.987753966045088e-05,
-      "loss": 0.1899,
+      "loss": 0.0891,
       "step": 66
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.987568420076074e-05,
-      "loss": 0.2437,
+      "loss": 0.0889,
       "step": 67
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9873828741070602e-05,
-      "loss": 0.1405,
+      "loss": 0.0262,
       "step": 68
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9871973281380464e-05,
-      "loss": 0.1918,
+      "loss": 0.1396,
       "step": 69
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9870117821690326e-05,
-      "loss": 0.1902,
+      "loss": 0.0289,
       "step": 70
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9868262362000188e-05,
-      "loss": 0.2397,
+      "loss": 0.1448,
       "step": 71
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.986640690231005e-05,
-      "loss": 0.2399,
+      "loss": 0.2549,
       "step": 72
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.986455144261991e-05,
-      "loss": 0.1598,
+      "loss": 0.1409,
       "step": 73
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9862695982929773e-05,
-      "loss": 0.1514,
+      "loss": 0.0871,
       "step": 74
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9860840523239635e-05,
-      "loss": 0.2817,
+      "loss": 0.0849,
       "step": 75
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9858985063549497e-05,
-      "loss": 0.1347,
+      "loss": 0.0267,
       "step": 76
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.985712960385936e-05,
-      "loss": 0.0976,
+      "loss": 0.0782,
       "step": 77
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9855274144169217e-05,
-      "loss": 0.1868,
+      "loss": 0.0278,
       "step": 78
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9853418684479083e-05,
-      "loss": 0.1408,
+      "loss": 0.079,
       "step": 79
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9851563224788945e-05,
-      "loss": 0.195,
+      "loss": 0.1958,
       "step": 80
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9849707765098803e-05,
-      "loss": 0.1992,
+      "loss": 0.0887,
       "step": 81
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9847852305408665e-05,
-      "loss": 0.185,
+      "loss": 0.0243,
       "step": 82
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.984599684571853e-05,
-      "loss": 0.1391,
+      "loss": 0.0219,
       "step": 83
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9844141386028392e-05,
-      "loss": 0.1921,
+      "loss": 0.2795,
       "step": 84
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.984228592633825e-05,
-      "loss": 0.0905,
+      "loss": 0.2091,
       "step": 85
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9840430466648112e-05,
-      "loss": 0.1446,
+      "loss": 0.2703,
       "step": 86
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9838575006957978e-05,
-      "loss": 0.2937,
+      "loss": 0.0845,
       "step": 87
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9836719547267836e-05,
-      "loss": 0.0917,
+      "loss": 0.1998,
       "step": 88
     },
     {
       "epoch": 0.02,
       "learning_rate": 1.9834864087577698e-05,
-      "loss": 0.1356,
+      "loss": 0.084,
       "step": 89
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.983300862788756e-05,
-      "loss": 0.0898,
+      "loss": 0.0957,
       "step": 90
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9831153168197425e-05,
-      "loss": 0.0892,
+      "loss": 0.0352,
       "step": 91
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9829297708507284e-05,
-      "loss": 0.035,
+      "loss": 0.1384,
       "step": 92
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9827442248817145e-05,
-      "loss": 0.0877,
+      "loss": 0.1998,
       "step": 93
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9825586789127007e-05,
-      "loss": 0.2562,
+      "loss": 0.0249,
       "step": 94
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.982373132943687e-05,
-      "loss": 0.1944,
+      "loss": 0.0863,
       "step": 95
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.982187586974673e-05,
-      "loss": 0.1342,
+      "loss": 0.0846,
       "step": 96
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9820020410056593e-05,
-      "loss": 0.0841,
+      "loss": 0.0829,
       "step": 97
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9818164950366455e-05,
-      "loss": 0.3032,
+      "loss": 0.1405,
       "step": 98
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9816309490676317e-05,
-      "loss": 0.1462,
+      "loss": 0.1946,
       "step": 99
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.981445403098618e-05,
-      "loss": 0.139,
+      "loss": 0.1475,
       "step": 100
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.981259857129604e-05,
-      "loss": 0.0901,
+      "loss": 0.2533,
       "step": 101
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9810743111605902e-05,
-      "loss": 0.1938,
+      "loss": 0.027,
       "step": 102
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9808887651915764e-05,
-      "loss": 0.138,
+      "loss": 0.1396,
       "step": 103
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9807032192225626e-05,
-      "loss": 0.0392,
+      "loss": 0.1327,
       "step": 104
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9805176732535488e-05,
-      "loss": 0.1977,
+      "loss": 0.0332,
       "step": 105
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.980332127284535e-05,
-      "loss": 0.1946,
+      "loss": 0.2037,
       "step": 106
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.980146581315521e-05,
-      "loss": 0.2936,
+      "loss": 0.1393,
       "step": 107
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9799610353465073e-05,
-      "loss": 0.1395,
+      "loss": 0.1898,
       "step": 108
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9797754893774935e-05,
-      "loss": 0.0837,
+      "loss": 0.1315,
       "step": 109
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9795899434084794e-05,
-      "loss": 0.2466,
+      "loss": 0.0926,
       "step": 110
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.979404397439466e-05,
-      "loss": 0.1951,
+      "loss": 0.0362,
       "step": 111
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.979218851470452e-05,
-      "loss": 0.0896,
+      "loss": 0.2515,
       "step": 112
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9790333055014383e-05,
-      "loss": 0.1839,
+      "loss": 0.0879,
       "step": 113
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.978847759532424e-05,
-      "loss": 0.0901,
+      "loss": 0.082,
       "step": 114
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9786622135634107e-05,
-      "loss": 0.2419,
+      "loss": 0.1467,
       "step": 115
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.978476667594397e-05,
-      "loss": 0.1944,
+      "loss": 0.0854,
       "step": 116
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9782911216253827e-05,
-      "loss": 0.2416,
+      "loss": 0.0859,
       "step": 117
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.978105575656369e-05,
-      "loss": 0.2341,
+      "loss": 0.0269,
       "step": 118
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9779200296873554e-05,
-      "loss": 0.1424,
+      "loss": 0.1417,
       "step": 119
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9777344837183416e-05,
-      "loss": 0.1417,
+      "loss": 0.0839,
       "step": 120
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9775489377493274e-05,
-      "loss": 0.1892,
+      "loss": 0.0821,
       "step": 121
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9773633917803136e-05,
-      "loss": 0.3231,
+      "loss": 0.199,
       "step": 122
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9771778458113e-05,
-      "loss": 0.1028,
+      "loss": 0.1366,
       "step": 123
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.976992299842286e-05,
-      "loss": 0.1439,
+      "loss": 0.1965,
       "step": 124
     },
     {
       "epoch": 0.03,
       "learning_rate": 1.9768067538732722e-05,
-      "loss": 0.2259,
+      "loss": 0.1412,
       "step": 125
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9766212079042584e-05,
-      "loss": 0.1401,
+      "loss": 0.086,
       "step": 126
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9764356619352446e-05,
-      "loss": 0.1904,
+      "loss": 0.0835,
       "step": 127
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9762501159662308e-05,
-      "loss": 0.0931,
+      "loss": 0.1986,
       "step": 128
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.976064569997217e-05,
-      "loss": 0.0438,
+      "loss": 0.3112,
       "step": 129
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.975879024028203e-05,
-      "loss": 0.1865,
+      "loss": 0.0838,
       "step": 130
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9756934780591893e-05,
-      "loss": 0.1949,
+      "loss": 0.1383,
       "step": 131
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9755079320901755e-05,
-      "loss": 0.2473,
+      "loss": 0.1411,
       "step": 132
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9753223861211617e-05,
-      "loss": 0.1929,
+      "loss": 0.1392,
       "step": 133
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.975136840152148e-05,
-      "loss": 0.2499,
+      "loss": 0.0481,
       "step": 134
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.974951294183134e-05,
-      "loss": 0.0857,
+      "loss": 0.0395,
       "step": 135
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9747657482141202e-05,
-      "loss": 0.2468,
+      "loss": 0.0305,
       "step": 136
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9745802022451064e-05,
-      "loss": 0.2912,
+      "loss": 0.2518,
       "step": 137
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9743946562760926e-05,
-      "loss": 0.197,
+      "loss": 0.141,
       "step": 138
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9742091103070788e-05,
-      "loss": 0.138,
+      "loss": 0.0831,
       "step": 139
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.974023564338065e-05,
-      "loss": 0.3418,
+      "loss": 0.0234,
       "step": 140
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9738380183690512e-05,
-      "loss": 0.2883,
+      "loss": 0.0825,
       "step": 141
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9736524724000374e-05,
-      "loss": 0.1477,
+      "loss": 0.2079,
       "step": 142
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9734669264310236e-05,
-      "loss": 0.1883,
+      "loss": 0.0238,
       "step": 143
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9732813804620097e-05,
-      "loss": 0.1863,
+      "loss": 0.0783,
       "step": 144
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.973095834492996e-05,
-      "loss": 0.1911,
+      "loss": 0.1422,
       "step": 145
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9729102885239818e-05,
-      "loss": 0.0675,
+      "loss": 0.084,
       "step": 146
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9727247425549683e-05,
-      "loss": 0.1831,
+      "loss": 0.199,
       "step": 147
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9725391965859545e-05,
-      "loss": 0.2313,
+      "loss": 0.1505,
       "step": 148
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9723536506169403e-05,
-      "loss": 0.0977,
+      "loss": 0.0249,
       "step": 149
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9721681046479265e-05,
-      "loss": 0.096,
+      "loss": 0.2037,
       "step": 150
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.971982558678913e-05,
-      "loss": 0.0917,
+      "loss": 0.0863,
       "step": 151
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9717970127098992e-05,
-      "loss": 0.3411,
+      "loss": 0.0809,
       "step": 152
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.971611466740885e-05,
-      "loss": 0.245,
+      "loss": 0.0252,
       "step": 153
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9714259207718713e-05,
-      "loss": 0.1386,
+      "loss": 0.1415,
       "step": 154
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9712403748028575e-05,
-      "loss": 0.3496,
+      "loss": 0.0238,
       "step": 155
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9710548288338437e-05,
-      "loss": 0.1908,
+      "loss": 0.1374,
       "step": 156
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.97086928286483e-05,
-      "loss": 0.0387,
+      "loss": 0.1432,
       "step": 157
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.970683736895816e-05,
-      "loss": 0.1928,
+      "loss": 0.2,
       "step": 158
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9704981909268022e-05,
-      "loss": 0.1866,
+      "loss": 0.0249,
       "step": 159
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9703126449577884e-05,
-      "loss": 0.1404,
+      "loss": 0.259,
       "step": 160
     },
     {
       "epoch": 0.04,
       "learning_rate": 1.9701270989887746e-05,
-      "loss": 0.1443,
+      "loss": 0.1946,
       "step": 161
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9699415530197608e-05,
-      "loss": 0.1382,
+      "loss": 0.1394,
       "step": 162
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.969756007050747e-05,
-      "loss": 0.1425,
+      "loss": 0.0303,
       "step": 163
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.969570461081733e-05,
-      "loss": 0.0933,
+      "loss": 0.0285,
       "step": 164
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9693849151127193e-05,
-      "loss": 0.0923,
+      "loss": 0.1392,
       "step": 165
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9691993691437055e-05,
-      "loss": 0.1375,
+      "loss": 0.0851,
       "step": 166
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9690138231746917e-05,
-      "loss": 0.1387,
+      "loss": 0.1407,
       "step": 167
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.968828277205678e-05,
-      "loss": 0.1396,
+      "loss": 0.0275,
       "step": 168
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.968642731236664e-05,
-      "loss": 0.1879,
+      "loss": 0.1409,
       "step": 169
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9684571852676503e-05,
-      "loss": 0.2436,
+      "loss": 0.0238,
       "step": 170
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9682716392986365e-05,
-      "loss": 0.3052,
+      "loss": 0.1473,
       "step": 171
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9680860933296226e-05,
-      "loss": 0.1949,
+      "loss": 0.2058,
       "step": 172
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9679005473606088e-05,
-      "loss": 0.1376,
+      "loss": 0.0235,
       "step": 173
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.967715001391595e-05,
-      "loss": 0.3447,
+      "loss": 0.0222,
       "step": 174
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.967529455422581e-05,
-      "loss": 0.0924,
+      "loss": 0.1317,
       "step": 175
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9673439094535674e-05,
-      "loss": 0.1402,
+      "loss": 0.2069,
       "step": 176
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9671583634845536e-05,
-      "loss": 0.0389,
+      "loss": 0.0828,
       "step": 177
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9669728175155394e-05,
-      "loss": 0.3385,
+      "loss": 0.0805,
       "step": 178
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9667872715465256e-05,
-      "loss": 0.1414,
+      "loss": 0.1415,
       "step": 179
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.966601725577512e-05,
-      "loss": 0.0916,
+      "loss": 0.2018,
       "step": 180
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9664161796084983e-05,
-      "loss": 0.0944,
+      "loss": 0.1986,
       "step": 181
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9662306336394842e-05,
-      "loss": 0.0949,
+      "loss": 0.0815,
       "step": 182
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9660450876704704e-05,
-      "loss": 0.0426,
+      "loss": 0.0814,
       "step": 183
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.965859541701457e-05,
-      "loss": 0.1382,
+      "loss": 0.0827,
       "step": 184
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9656739957324427e-05,
-      "loss": 0.3478,
+      "loss": 0.2488,
       "step": 185
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.965488449763429e-05,
-      "loss": 0.1406,
+      "loss": 0.0832,
       "step": 186
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.965302903794415e-05,
-      "loss": 0.1976,
+      "loss": 0.1401,
       "step": 187
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9651173578254016e-05,
-      "loss": 0.1892,
+      "loss": 0.0858,
       "step": 188
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9649318118563875e-05,
-      "loss": 0.1388,
+      "loss": 0.3002,
       "step": 189
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9647462658873737e-05,
-      "loss": 0.2969,
+      "loss": 0.0343,
       "step": 190
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.96456071991836e-05,
-      "loss": 0.1421,
+      "loss": 0.0964,
       "step": 191
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.964375173949346e-05,
-      "loss": 0.0872,
+      "loss": 0.2369,
       "step": 192
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9641896279803322e-05,
-      "loss": 0.2473,
+      "loss": 0.2933,
       "step": 193
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9640040820113184e-05,
-      "loss": 0.2842,
+      "loss": 0.191,
       "step": 194
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9638185360423046e-05,
-      "loss": 0.145,
+      "loss": 0.0925,
       "step": 195
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.9636329900732908e-05,
-      "loss": 0.1424,
+      "loss": 0.0966,
       "step": 196
     },
     {
       "epoch": 0.05,
       "learning_rate": 1.963447444104277e-05,
-      "loss": 0.0926,
+      "loss": 0.2799,
       "step": 197
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.963261898135263e-05,
-      "loss": 0.0899,
+      "loss": 0.232,
       "step": 198
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9630763521662494e-05,
-      "loss": 0.0908,
+      "loss": 0.0541,
       "step": 199
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9628908061972355e-05,
-      "loss": 0.1411,
+      "loss": 0.1863,
       "step": 200
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9627052602282217e-05,
-      "loss": 0.1395,
+      "loss": 0.0928,
       "step": 201
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.962519714259208e-05,
-      "loss": 0.139,
+      "loss": 0.0903,
       "step": 202
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.962334168290194e-05,
-      "loss": 0.0878,
+      "loss": 0.1384,
       "step": 203
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9621486223211803e-05,
-      "loss": 0.2413,
+      "loss": 0.244,
       "step": 204
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9619630763521665e-05,
-      "loss": 0.1858,
+      "loss": 0.0879,
       "step": 205
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9617775303831527e-05,
-      "loss": 0.2483,
+      "loss": 0.0877,
       "step": 206
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9615919844141385e-05,
-      "loss": 0.2944,
+      "loss": 0.1991,
       "step": 207
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.961406438445125e-05,
-      "loss": 0.0846,
+      "loss": 0.1917,
       "step": 208
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9612208924761112e-05,
-      "loss": 0.138,
+      "loss": 0.1384,
       "step": 209
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9610353465070974e-05,
-      "loss": 0.2424,
+      "loss": 0.1389,
       "step": 210
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9608498005380833e-05,
-      "loss": 0.1401,
+      "loss": 0.0275,
       "step": 211
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9606642545690698e-05,
-      "loss": 0.1352,
+      "loss": 0.2047,
       "step": 212
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.960478708600056e-05,
-      "loss": 0.1398,
+      "loss": 0.1391,
       "step": 213
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9602931626310418e-05,
-      "loss": 0.1415,
+      "loss": 0.0856,
       "step": 214
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.960107616662028e-05,
-      "loss": 0.2496,
+      "loss": 0.2533,
       "step": 215
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9599220706930145e-05,
-      "loss": 0.1401,
+      "loss": 0.0301,
       "step": 216
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9597365247240007e-05,
-      "loss": 0.1362,
+      "loss": 0.1445,
       "step": 217
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9595509787549866e-05,
-      "loss": 0.0882,
+      "loss": 0.031,
       "step": 218
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9593654327859728e-05,
-      "loss": 0.039,
+      "loss": 0.2514,
       "step": 219
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9591798868169593e-05,
-      "loss": 0.1373,
+      "loss": 0.0895,
       "step": 220
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.958994340847945e-05,
-      "loss": 0.243,
+      "loss": 0.1403,
       "step": 221
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9588087948789313e-05,
-      "loss": 0.1943,
+      "loss": 0.1405,
       "step": 222
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9586232489099175e-05,
-      "loss": 0.134,
+      "loss": 0.1963,
       "step": 223
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9584377029409037e-05,
-      "loss": 0.0366,
+      "loss": 0.0823,
       "step": 224
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.95825215697189e-05,
-      "loss": 0.3406,
+      "loss": 0.0343,
       "step": 225
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.958066611002876e-05,
-      "loss": 0.144,
+      "loss": 0.0863,
       "step": 226
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9578810650338623e-05,
-      "loss": 0.0371,
+      "loss": 0.0283,
       "step": 227
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9576955190648484e-05,
-      "loss": 0.0365,
+      "loss": 0.1324,
       "step": 228
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9575099730958346e-05,
-      "loss": 0.2976,
+      "loss": 0.0829,
       "step": 229
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9573244271268208e-05,
-      "loss": 0.1938,
+      "loss": 0.0837,
       "step": 230
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.957138881157807e-05,
-      "loss": 0.191,
+      "loss": 0.1367,
       "step": 231
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9569533351887932e-05,
-      "loss": 0.0866,
+      "loss": 0.0194,
       "step": 232
     },
     {
       "epoch": 0.06,
       "learning_rate": 1.9567677892197794e-05,
-      "loss": 0.0853,
+      "loss": 0.0202,
       "step": 233
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9565822432507656e-05,
-      "loss": 0.3506,
+      "loss": 0.1479,
       "step": 234
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9563966972817518e-05,
-      "loss": 0.0912,
+      "loss": 0.1475,
       "step": 235
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.956211151312738e-05,
-      "loss": 0.0365,
+      "loss": 0.0812,
       "step": 236
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.956025605343724e-05,
-      "loss": 0.1414,
+      "loss": 0.2642,
       "step": 237
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9558400593747103e-05,
-      "loss": 0.1358,
+      "loss": 0.0828,
       "step": 238
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9556545134056965e-05,
-      "loss": 0.2391,
+      "loss": 0.1329,
       "step": 239
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9554689674366827e-05,
-      "loss": 0.0348,
+      "loss": 0.0255,
       "step": 240
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.955283421467669e-05,
-      "loss": 0.1853,
+      "loss": 0.0828,
       "step": 241
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.955097875498655e-05,
-      "loss": 0.1939,
+      "loss": 0.136,
       "step": 242
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.954912329529641e-05,
-      "loss": 0.1386,
+      "loss": 0.1932,
       "step": 243
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9547267835606274e-05,
-      "loss": 0.1906,
+      "loss": 0.084,
       "step": 244
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9545412375916136e-05,
-      "loss": 0.1334,
+      "loss": 0.1871,
       "step": 245
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9543556916225995e-05,
-      "loss": 0.142,
+      "loss": 0.1388,
       "step": 246
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9541701456535857e-05,
-      "loss": 0.0338,
+      "loss": 0.2472,
       "step": 247
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9539845996845722e-05,
-      "loss": 0.1874,
+      "loss": 0.0967,
       "step": 248
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9537990537155584e-05,
-      "loss": 0.1414,
+      "loss": 0.0963,
       "step": 249
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9536135077465442e-05,
-      "loss": 0.0851,
+      "loss": 0.0408,
       "step": 250
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9534279617775304e-05,
-      "loss": 0.3014,
+      "loss": 0.0311,
       "step": 251
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.953242415808517e-05,
-      "loss": 0.1395,
+      "loss": 0.1374,
       "step": 252
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9530568698395028e-05,
-      "loss": 0.0832,
+      "loss": 0.1451,
       "step": 253
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.952871323870489e-05,
-      "loss": 0.0859,
+      "loss": 0.0808,
       "step": 254
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.952685777901475e-05,
-      "loss": 0.0871,
+      "loss": 0.081,
       "step": 255
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9525002319324617e-05,
-      "loss": 0.0886,
+      "loss": 0.0178,
       "step": 256
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9523146859634475e-05,
-      "loss": 0.0854,
+      "loss": 0.1997,
       "step": 257
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9521291399944337e-05,
-      "loss": 0.0857,
+      "loss": 0.2098,
       "step": 258
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.95194359402542e-05,
-      "loss": 0.1967,
+      "loss": 0.1428,
       "step": 259
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.951758048056406e-05,
-      "loss": 0.0276,
+      "loss": 0.1441,
       "step": 260
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9515725020873923e-05,
-      "loss": 0.0816,
+      "loss": 0.1428,
       "step": 261
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9513869561183785e-05,
-      "loss": 0.1972,
+      "loss": 0.0807,
       "step": 262
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9512014101493646e-05,
-      "loss": 0.3668,
+      "loss": 0.1968,
       "step": 263
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.951015864180351e-05,
-      "loss": 0.2423,
+      "loss": 0.0835,
       "step": 264
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.950830318211337e-05,
-      "loss": 0.1929,
+      "loss": 0.0257,
       "step": 265
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9506447722423232e-05,
-      "loss": 0.3961,
+      "loss": 0.1355,
       "step": 266
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9504592262733094e-05,
-      "loss": 0.3523,
+      "loss": 0.1359,
       "step": 267
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9502736803042956e-05,
-      "loss": 0.3257,
+      "loss": 0.0257,
       "step": 268
     },
     {
       "epoch": 0.07,
       "learning_rate": 1.9500881343352818e-05,
-      "loss": 0.2013,
+      "loss": 0.0819,
       "step": 269
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.949902588366268e-05,
-      "loss": 0.1229,
+      "loss": 0.0819,
       "step": 270
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.949717042397254e-05,
-      "loss": 0.1518,
+      "loss": 0.0259,
       "step": 271
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9495314964282403e-05,
-      "loss": 0.0525,
+      "loss": 0.0837,
       "step": 272
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9493459504592265e-05,
-      "loss": 0.1402,
+      "loss": 0.0228,
       "step": 273
     },
     {
@@ -1654,22129 +1654,22129 @@
     {
       "epoch": 0.08,
       "learning_rate": 1.9489748585211986e-05,
-      "loss": 0.1405,
+      "loss": 0.3861,
       "step": 275
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.948789312552185e-05,
-      "loss": 0.0397,
+      "loss": 0.0208,
       "step": 276
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9486037665831713e-05,
-      "loss": 0.1415,
+      "loss": 0.0793,
       "step": 277
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9484182206141575e-05,
-      "loss": 0.1968,
+      "loss": 0.2038,
       "step": 278
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9482326746451433e-05,
-      "loss": 0.3479,
+      "loss": 0.0774,
       "step": 279
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9480471286761298e-05,
-      "loss": 0.3021,
+      "loss": 0.086,
       "step": 280
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.947861582707116e-05,
-      "loss": 0.092,
+      "loss": 0.204,
       "step": 281
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.947676036738102e-05,
-      "loss": 0.1447,
+      "loss": 0.1422,
       "step": 282
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.947490490769088e-05,
-      "loss": 0.0894,
+      "loss": 0.0796,
       "step": 283
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9473049448000746e-05,
-      "loss": 0.041,
+      "loss": 0.0774,
       "step": 284
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9471193988310608e-05,
-      "loss": 0.09,
+      "loss": 0.1394,
       "step": 285
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9469338528620466e-05,
-      "loss": 0.1388,
+      "loss": 0.083,
       "step": 286
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9467483068930328e-05,
-      "loss": 0.2469,
+      "loss": 0.1345,
       "step": 287
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9465627609240193e-05,
-      "loss": 0.1395,
+      "loss": 0.1391,
       "step": 288
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9463772149550052e-05,
-      "loss": 0.1912,
+      "loss": 0.0817,
       "step": 289
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9461916689859914e-05,
-      "loss": 0.1399,
+      "loss": 0.2609,
       "step": 290
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9460061230169775e-05,
-      "loss": 0.1385,
+      "loss": 0.0792,
       "step": 291
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9458205770479637e-05,
-      "loss": 0.0866,
+      "loss": 0.0853,
       "step": 292
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.94563503107895e-05,
-      "loss": 0.0873,
+      "loss": 0.3025,
       "step": 293
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.945449485109936e-05,
-      "loss": 0.1398,
+      "loss": 0.1356,
       "step": 294
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9452639391409223e-05,
-      "loss": 0.2439,
+      "loss": 0.1357,
       "step": 295
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9450783931719085e-05,
-      "loss": 0.1932,
+      "loss": 0.0336,
       "step": 296
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9448928472028947e-05,
-      "loss": 0.2422,
+      "loss": 0.2373,
       "step": 297
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.944707301233881e-05,
-      "loss": 0.193,
+      "loss": 0.1382,
       "step": 298
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.944521755264867e-05,
-      "loss": 0.0856,
+      "loss": 0.1421,
       "step": 299
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9443362092958532e-05,
-      "loss": 0.1951,
+      "loss": 0.0885,
       "step": 300
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9441506633268394e-05,
-      "loss": 0.1401,
+      "loss": 0.1873,
       "step": 301
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9439651173578256e-05,
-      "loss": 0.2402,
+      "loss": 0.0856,
       "step": 302
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9437795713888118e-05,
-      "loss": 0.29,
+      "loss": 0.1361,
       "step": 303
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.9435940254197976e-05,
-      "loss": 0.1952,
+      "loss": 0.1388,
       "step": 304
     },
     {
       "epoch": 0.08,
       "learning_rate": 1.943408479450784e-05,
-      "loss": 0.0528,
+      "loss": 0.2469,
       "step": 305
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9432229334817704e-05,
-      "loss": 0.3256,
+      "loss": 0.0384,
       "step": 306
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9430373875127565e-05,
-      "loss": 0.1917,
+      "loss": 0.0886,
       "step": 307
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9428518415437424e-05,
-      "loss": 0.0602,
+      "loss": 0.194,
       "step": 308
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.942666295574729e-05,
-      "loss": 0.2792,
+      "loss": 0.1404,
       "step": 309
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.942480749605715e-05,
-      "loss": 0.1015,
+      "loss": 0.0335,
       "step": 310
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.942295203636701e-05,
-      "loss": 0.1907,
+      "loss": 0.1412,
       "step": 311
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.942109657667687e-05,
-      "loss": 0.051,
+      "loss": 0.1349,
       "step": 312
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9419241116986737e-05,
-      "loss": 0.1881,
+      "loss": 0.3122,
       "step": 313
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.94173856572966e-05,
-      "loss": 0.287,
+      "loss": 0.2437,
       "step": 314
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9415530197606457e-05,
-      "loss": 0.088,
+      "loss": 0.0296,
       "step": 315
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.941367473791632e-05,
-      "loss": 0.1828,
+      "loss": 0.1932,
       "step": 316
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9411819278226184e-05,
-      "loss": 0.0385,
+      "loss": 0.0832,
       "step": 317
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9409963818536043e-05,
-      "loss": 0.0871,
+      "loss": 0.1905,
       "step": 318
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9408108358845904e-05,
-      "loss": 0.1365,
+      "loss": 0.0845,
       "step": 319
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9406252899155766e-05,
-      "loss": 0.1934,
+      "loss": 0.0825,
       "step": 320
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9404397439465628e-05,
-      "loss": 0.0341,
+      "loss": 0.1395,
       "step": 321
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.940254197977549e-05,
-      "loss": 0.2469,
+      "loss": 0.0323,
       "step": 322
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9400686520085352e-05,
-      "loss": 0.1381,
+      "loss": 0.3576,
       "step": 323
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9398831060395214e-05,
-      "loss": 0.188,
+      "loss": 0.1855,
       "step": 324
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9396975600705076e-05,
-      "loss": 0.4072,
+      "loss": 0.1359,
       "step": 325
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9395120141014938e-05,
-      "loss": 0.1911,
+      "loss": 0.0295,
       "step": 326
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.93932646813248e-05,
-      "loss": 0.134,
+      "loss": 0.0875,
       "step": 327
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.939140922163466e-05,
-      "loss": 0.1402,
+      "loss": 0.0307,
       "step": 328
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9389553761944523e-05,
-      "loss": 0.0329,
+      "loss": 0.1959,
       "step": 329
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9387698302254385e-05,
-      "loss": 0.1342,
+      "loss": 0.0283,
       "step": 330
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9385842842564247e-05,
-      "loss": 0.2953,
+      "loss": 0.1392,
       "step": 331
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.938398738287411e-05,
-      "loss": 0.2454,
+      "loss": 0.084,
       "step": 332
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.938213192318397e-05,
-      "loss": 0.2457,
+      "loss": 0.196,
       "step": 333
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9380276463493833e-05,
-      "loss": 0.0879,
+      "loss": 0.0838,
       "step": 334
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9378421003803694e-05,
-      "loss": 0.1413,
+      "loss": 0.0244,
       "step": 335
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9376565544113556e-05,
-      "loss": 0.1834,
+      "loss": 0.1398,
       "step": 336
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9374710084423418e-05,
-      "loss": 0.2467,
+      "loss": 0.0804,
       "step": 337
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.937285462473328e-05,
-      "loss": 0.1418,
+      "loss": 0.081,
       "step": 338
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9370999165043142e-05,
-      "loss": 0.0913,
+      "loss": 0.1331,
       "step": 339
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9369143705353e-05,
-      "loss": 0.1504,
+      "loss": 0.0856,
       "step": 340
     },
     {
       "epoch": 0.09,
       "learning_rate": 1.9367288245662866e-05,
-      "loss": 0.2328,
+      "loss": 0.2035,
       "step": 341
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9365432785972727e-05,
-      "loss": 0.2432,
+      "loss": 0.0209,
       "step": 342
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.936357732628259e-05,
-      "loss": 0.096,
+      "loss": 0.08,
       "step": 343
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9361721866592448e-05,
-      "loss": 0.1449,
+      "loss": 0.0772,
       "step": 344
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9359866406902313e-05,
-      "loss": 0.3872,
+      "loss": 0.1405,
       "step": 345
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9358010947212175e-05,
-      "loss": 0.0894,
+      "loss": 0.1388,
       "step": 346
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9356155487522033e-05,
-      "loss": 0.1439,
+      "loss": 0.1427,
       "step": 347
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9354300027831895e-05,
-      "loss": 0.1417,
+      "loss": 0.147,
       "step": 348
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.935244456814176e-05,
-      "loss": 0.0974,
+      "loss": 0.0826,
       "step": 349
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.935058910845162e-05,
-      "loss": 0.1408,
+      "loss": 0.196,
       "step": 350
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.934873364876148e-05,
-      "loss": 0.091,
+      "loss": 0.0233,
       "step": 351
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9346878189071343e-05,
-      "loss": 0.091,
+      "loss": 0.1418,
       "step": 352
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9345022729381208e-05,
-      "loss": 0.186,
+      "loss": 0.1372,
       "step": 353
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9343167269691067e-05,
-      "loss": 0.0882,
+      "loss": 0.205,
       "step": 354
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.934131181000093e-05,
-      "loss": 0.1396,
+      "loss": 0.0833,
       "step": 355
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.933945635031079e-05,
-      "loss": 0.1378,
+      "loss": 0.1365,
       "step": 356
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9337600890620652e-05,
-      "loss": 0.1391,
+      "loss": 0.2585,
       "step": 357
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9335745430930514e-05,
-      "loss": 0.1942,
+      "loss": 0.1371,
       "step": 358
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9333889971240376e-05,
-      "loss": 0.138,
+      "loss": 0.0329,
       "step": 359
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9332034511550238e-05,
-      "loss": 0.1388,
+      "loss": 0.0855,
       "step": 360
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.93301790518601e-05,
-      "loss": 0.0872,
+      "loss": 0.0914,
       "step": 361
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.932832359216996e-05,
-      "loss": 0.0314,
+      "loss": 0.19,
       "step": 362
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9326468132479823e-05,
-      "loss": 0.0866,
+      "loss": 0.1398,
       "step": 363
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9324612672789685e-05,
-      "loss": 0.2481,
+      "loss": 0.0865,
       "step": 364
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9322757213099547e-05,
-      "loss": 0.1924,
+      "loss": 0.0889,
       "step": 365
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.932090175340941e-05,
-      "loss": 0.1986,
+      "loss": 0.0877,
       "step": 366
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.931904629371927e-05,
-      "loss": 0.1896,
+      "loss": 0.0865,
       "step": 367
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9317190834029133e-05,
-      "loss": 0.029,
+      "loss": 0.0287,
       "step": 368
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9315335374338995e-05,
-      "loss": 0.0817,
+      "loss": 0.0864,
       "step": 369
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9313479914648856e-05,
-      "loss": 0.4144,
+      "loss": 0.1989,
       "step": 370
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.931162445495872e-05,
-      "loss": 0.1952,
+      "loss": 0.0235,
       "step": 371
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9309768995268577e-05,
-      "loss": 0.0833,
+      "loss": 0.0776,
       "step": 372
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9307913535578442e-05,
-      "loss": 0.1367,
+      "loss": 0.2019,
       "step": 373
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9306058075888304e-05,
-      "loss": 0.0886,
+      "loss": 0.146,
       "step": 374
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9304202616198166e-05,
-      "loss": 0.1398,
+      "loss": 0.141,
       "step": 375
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.9302347156508024e-05,
-      "loss": 0.1363,
+      "loss": 0.0835,
       "step": 376
     },
     {
       "epoch": 0.1,
       "learning_rate": 1.930049169681789e-05,
-      "loss": 0.0857,
+      "loss": 0.2034,
       "step": 377
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.929863623712775e-05,
-      "loss": 0.3012,
+      "loss": 0.0816,
       "step": 378
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.929678077743761e-05,
-      "loss": 0.1963,
+      "loss": 0.0215,
       "step": 379
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9294925317747472e-05,
-      "loss": 0.1928,
+      "loss": 0.0196,
       "step": 380
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9293069858057337e-05,
-      "loss": 0.083,
+      "loss": 0.0197,
       "step": 381
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.92912143983672e-05,
-      "loss": 0.1919,
+      "loss": 0.0191,
       "step": 382
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9289358938677057e-05,
-      "loss": 0.1903,
+      "loss": 0.0803,
       "step": 383
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.928750347898692e-05,
-      "loss": 0.2431,
+      "loss": 0.1371,
       "step": 384
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9285648019296785e-05,
-      "loss": 0.2363,
+      "loss": 0.1443,
       "step": 385
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9283792559606643e-05,
-      "loss": 0.0892,
+      "loss": 0.0777,
       "step": 386
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9281937099916505e-05,
-      "loss": 0.2401,
+      "loss": 0.0801,
       "step": 387
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9280081640226367e-05,
-      "loss": 0.1873,
+      "loss": 0.2054,
       "step": 388
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.927822618053623e-05,
-      "loss": 0.1894,
+      "loss": 0.2127,
       "step": 389
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.927637072084609e-05,
-      "loss": 0.2956,
+      "loss": 0.207,
       "step": 390
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9274515261155952e-05,
-      "loss": 0.2844,
+      "loss": 0.0776,
       "step": 391
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9272659801465814e-05,
-      "loss": 0.0517,
+      "loss": 0.1472,
       "step": 392
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9270804341775676e-05,
-      "loss": 0.2418,
+      "loss": 0.0793,
       "step": 393
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9268948882085538e-05,
-      "loss": 0.1433,
+      "loss": 0.2568,
       "step": 394
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.92670934223954e-05,
-      "loss": 0.1002,
+      "loss": 0.2034,
       "step": 395
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9265237962705262e-05,
-      "loss": 0.054,
+      "loss": 0.029,
       "step": 396
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9263382503015124e-05,
-      "loss": 0.277,
+      "loss": 0.1347,
       "step": 397
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9261527043324985e-05,
-      "loss": 0.2301,
+      "loss": 0.2497,
       "step": 398
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9259671583634847e-05,
-      "loss": 0.2873,
+      "loss": 0.0887,
       "step": 399
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.925781612394471e-05,
-      "loss": 0.1898,
+      "loss": 0.0892,
       "step": 400
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.925596066425457e-05,
-      "loss": 0.2371,
+      "loss": 0.0893,
       "step": 401
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9254105204564433e-05,
-      "loss": 0.1912,
+      "loss": 0.1396,
       "step": 402
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9252249744874295e-05,
-      "loss": 0.1461,
+      "loss": 0.193,
       "step": 403
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9250394285184157e-05,
-      "loss": 0.1888,
+      "loss": 0.1393,
       "step": 404
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.924853882549402e-05,
-      "loss": 0.1899,
+      "loss": 0.2031,
       "step": 405
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.924668336580388e-05,
-      "loss": 0.1404,
+      "loss": 0.0374,
       "step": 406
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9244827906113742e-05,
-      "loss": 0.1874,
+      "loss": 0.2397,
       "step": 407
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.92429724464236e-05,
-      "loss": 0.0967,
+      "loss": 0.0364,
       "step": 408
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9241116986733466e-05,
-      "loss": 0.1883,
+      "loss": 0.193,
       "step": 409
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9239261527043328e-05,
-      "loss": 0.1445,
+      "loss": 0.0868,
       "step": 410
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.923740606735319e-05,
-      "loss": 0.0936,
+      "loss": 0.1367,
       "step": 411
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9235550607663048e-05,
-      "loss": 0.2842,
+      "loss": 0.1419,
       "step": 412
     },
     {
       "epoch": 0.11,
       "learning_rate": 1.9233695147972914e-05,
-      "loss": 0.1389,
+      "loss": 0.0887,
       "step": 413
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9231839688282775e-05,
-      "loss": 0.1387,
+      "loss": 0.1436,
       "step": 414
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9229984228592634e-05,
-      "loss": 0.1861,
+      "loss": 0.1434,
       "step": 415
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9228128768902496e-05,
-      "loss": 0.0415,
+      "loss": 0.0291,
       "step": 416
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.922627330921236e-05,
-      "loss": 0.242,
+      "loss": 0.0829,
       "step": 417
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.922441784952222e-05,
-      "loss": 0.0882,
+      "loss": 0.084,
       "step": 418
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.922256238983208e-05,
-      "loss": 0.1347,
+      "loss": 0.19,
       "step": 419
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9220706930141943e-05,
-      "loss": 0.0364,
+      "loss": 0.1374,
       "step": 420
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.921885147045181e-05,
-      "loss": 0.0871,
+      "loss": 0.0827,
       "step": 421
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9216996010761667e-05,
-      "loss": 0.0327,
+      "loss": 0.0246,
       "step": 422
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.921514055107153e-05,
-      "loss": 0.0861,
+      "loss": 0.0221,
       "step": 423
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.921328509138139e-05,
-      "loss": 0.246,
+      "loss": 0.0857,
       "step": 424
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9211429631691253e-05,
-      "loss": 0.1958,
+      "loss": 0.0211,
       "step": 425
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9209574172001114e-05,
-      "loss": 0.1416,
+      "loss": 0.0773,
       "step": 426
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9207718712310976e-05,
-      "loss": 0.0849,
+      "loss": 0.0181,
       "step": 427
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9205863252620838e-05,
-      "loss": 0.0863,
+      "loss": 0.1437,
       "step": 428
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.92040077929307e-05,
-      "loss": 0.1416,
+      "loss": 0.0172,
       "step": 429
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9202152333240562e-05,
-      "loss": 0.2511,
+      "loss": 0.1503,
       "step": 430
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9200296873550424e-05,
-      "loss": 0.3139,
+      "loss": 0.0158,
       "step": 431
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9198441413860286e-05,
-      "loss": 0.1917,
+      "loss": 0.0806,
       "step": 432
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9196585954170148e-05,
-      "loss": 0.1973,
+      "loss": 0.082,
       "step": 433
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.919473049448001e-05,
-      "loss": 0.2492,
+      "loss": 0.1462,
       "step": 434
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.919287503478987e-05,
-      "loss": 0.1365,
+      "loss": 0.2086,
       "step": 435
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9191019575099733e-05,
-      "loss": 0.2427,
+      "loss": 0.1496,
       "step": 436
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9189164115409595e-05,
-      "loss": 0.2454,
+      "loss": 0.2181,
       "step": 437
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9187308655719457e-05,
-      "loss": 0.247,
+      "loss": 0.2772,
       "step": 438
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.918545319602932e-05,
-      "loss": 0.2445,
+      "loss": 0.08,
       "step": 439
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.918359773633918e-05,
-      "loss": 0.2366,
+      "loss": 0.0756,
       "step": 440
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9181742276649043e-05,
-      "loss": 0.1432,
+      "loss": 0.0828,
       "step": 441
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9179886816958904e-05,
-      "loss": 0.3311,
+      "loss": 0.0187,
       "step": 442
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9178031357268766e-05,
-      "loss": 0.0451,
+      "loss": 0.0187,
       "step": 443
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9176175897578625e-05,
-      "loss": 0.1433,
+      "loss": 0.0828,
       "step": 444
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.917432043788849e-05,
-      "loss": 0.0985,
+      "loss": 0.1412,
       "step": 445
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9172464978198352e-05,
-      "loss": 0.1443,
+      "loss": 0.149,
       "step": 446
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.917060951850821e-05,
-      "loss": 0.0958,
+      "loss": 0.1994,
       "step": 447
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9168754058818072e-05,
-      "loss": 0.1876,
+      "loss": 0.2563,
       "step": 448
     },
     {
       "epoch": 0.12,
       "learning_rate": 1.9166898599127937e-05,
-      "loss": 0.1866,
+      "loss": 0.1463,
       "step": 449
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.91650431394378e-05,
-      "loss": 0.1443,
+      "loss": 0.2538,
       "step": 450
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9163187679747658e-05,
-      "loss": 0.0983,
+      "loss": 0.1389,
       "step": 451
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.916133222005752e-05,
-      "loss": 0.1423,
+      "loss": 0.3577,
       "step": 452
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9159476760367385e-05,
-      "loss": 0.1425,
+      "loss": 0.1918,
       "step": 453
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9157621300677243e-05,
-      "loss": 0.0936,
+      "loss": 0.1401,
       "step": 454
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9155765840987105e-05,
-      "loss": 0.1955,
+      "loss": 0.1887,
       "step": 455
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9153910381296967e-05,
-      "loss": 0.041,
+      "loss": 0.1904,
       "step": 456
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.915205492160683e-05,
-      "loss": 0.1926,
+      "loss": 0.1476,
       "step": 457
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.915019946191669e-05,
-      "loss": 0.2412,
+      "loss": 0.2793,
       "step": 458
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9148344002226553e-05,
-      "loss": 0.1935,
+      "loss": 0.1477,
       "step": 459
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9146488542536415e-05,
-      "loss": 0.1902,
+      "loss": 0.0625,
       "step": 460
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9144633082846277e-05,
-      "loss": 0.1383,
+      "loss": 0.147,
       "step": 461
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.914277762315614e-05,
-      "loss": 0.0881,
+      "loss": 0.1399,
       "step": 462
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9140922163466e-05,
-      "loss": 0.1937,
+      "loss": 0.1817,
       "step": 463
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9139066703775862e-05,
-      "loss": 0.1937,
+      "loss": 0.1813,
       "step": 464
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9137211244085724e-05,
-      "loss": 0.0353,
+      "loss": 0.0936,
       "step": 465
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9135355784395586e-05,
-      "loss": 0.1906,
+      "loss": 0.0861,
       "step": 466
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9133500324705448e-05,
-      "loss": 0.1459,
+      "loss": 0.0857,
       "step": 467
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.913164486501531e-05,
-      "loss": 0.085,
+      "loss": 0.0336,
       "step": 468
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.912978940532517e-05,
-      "loss": 0.1424,
+      "loss": 0.1398,
       "step": 469
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9127933945635033e-05,
-      "loss": 0.1934,
+      "loss": 0.0247,
       "step": 470
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9126078485944895e-05,
-      "loss": 0.2446,
+      "loss": 0.1389,
       "step": 471
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9124223026254757e-05,
-      "loss": 0.1325,
+      "loss": 0.1395,
       "step": 472
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9122367566564616e-05,
-      "loss": 0.0843,
+      "loss": 0.3881,
       "step": 473
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.912051210687448e-05,
-      "loss": 0.1952,
+      "loss": 0.2012,
       "step": 474
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9118656647184343e-05,
-      "loss": 0.1921,
+      "loss": 0.3235,
       "step": 475
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.91168011874942e-05,
-      "loss": 0.0866,
+      "loss": 0.0831,
       "step": 476
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9114945727804063e-05,
-      "loss": 0.1427,
+      "loss": 0.1402,
       "step": 477
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.911309026811393e-05,
-      "loss": 0.1986,
+      "loss": 0.0272,
       "step": 478
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.911123480842379e-05,
-      "loss": 0.1362,
+      "loss": 0.0817,
       "step": 479
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.910937934873365e-05,
-      "loss": 0.2977,
+      "loss": 0.2552,
       "step": 480
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.910752388904351e-05,
-      "loss": 0.034,
+      "loss": 0.0867,
       "step": 481
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9105668429353376e-05,
-      "loss": 0.2403,
+      "loss": 0.0815,
       "step": 482
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9103812969663234e-05,
-      "loss": 0.1939,
+      "loss": 0.1887,
       "step": 483
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9101957509973096e-05,
-      "loss": 0.0876,
+      "loss": 0.087,
       "step": 484
     },
     {
       "epoch": 0.13,
       "learning_rate": 1.9100102050282958e-05,
-      "loss": 0.239,
+      "loss": 0.2499,
       "step": 485
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.909824659059282e-05,
-      "loss": 0.3498,
+      "loss": 0.2489,
       "step": 486
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9096391130902682e-05,
-      "loss": 0.2911,
+      "loss": 0.0907,
       "step": 487
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9094535671212544e-05,
-      "loss": 0.1922,
+      "loss": 0.0381,
       "step": 488
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9092680211522406e-05,
-      "loss": 0.042,
+      "loss": 0.1401,
       "step": 489
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9090824751832267e-05,
-      "loss": 0.1435,
+      "loss": 0.0858,
       "step": 490
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.908896929214213e-05,
-      "loss": 0.1925,
+      "loss": 0.1414,
       "step": 491
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.908711383245199e-05,
-      "loss": 0.1946,
+      "loss": 0.1358,
       "step": 492
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9085258372761853e-05,
-      "loss": 0.0968,
+      "loss": 0.0862,
       "step": 493
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9083402913071715e-05,
-      "loss": 0.1868,
+      "loss": 0.0308,
       "step": 494
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9081547453381577e-05,
-      "loss": 0.2683,
+      "loss": 0.2465,
       "step": 495
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.907969199369144e-05,
-      "loss": 0.0469,
+      "loss": 0.0288,
       "step": 496
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.90778365340013e-05,
-      "loss": 0.194,
+      "loss": 0.0836,
       "step": 497
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9075981074311162e-05,
-      "loss": 0.1454,
+      "loss": 0.0844,
       "step": 498
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9074125614621024e-05,
-      "loss": 0.1919,
+      "loss": 0.0805,
       "step": 499
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9072270154930886e-05,
-      "loss": 0.1911,
+      "loss": 0.0809,
       "step": 500
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9070414695240748e-05,
-      "loss": 0.0943,
+      "loss": 0.0218,
       "step": 501
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.906855923555061e-05,
-      "loss": 0.1926,
+      "loss": 0.1386,
       "step": 502
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9066703775860472e-05,
-      "loss": 0.1877,
+      "loss": 0.1465,
       "step": 503
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9064848316170334e-05,
-      "loss": 0.0916,
+      "loss": 0.2663,
       "step": 504
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9062992856480192e-05,
-      "loss": 0.1861,
+      "loss": 0.08,
       "step": 505
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9061137396790057e-05,
-      "loss": 0.1874,
+      "loss": 0.1399,
       "step": 506
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.905928193709992e-05,
-      "loss": 0.0437,
+      "loss": 0.1423,
       "step": 507
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.905742647740978e-05,
-      "loss": 0.2888,
+      "loss": 0.1365,
       "step": 508
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.905557101771964e-05,
-      "loss": 0.1903,
+      "loss": 0.0852,
       "step": 509
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9053715558029505e-05,
-      "loss": 0.2934,
+      "loss": 0.261,
       "step": 510
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9051860098339367e-05,
-      "loss": 0.1397,
+      "loss": 0.1426,
       "step": 511
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9050004638649225e-05,
-      "loss": 0.2432,
+      "loss": 0.1384,
       "step": 512
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9048149178959087e-05,
-      "loss": 0.2878,
+      "loss": 0.026,
       "step": 513
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9046293719268952e-05,
-      "loss": 0.2813,
+      "loss": 0.3005,
       "step": 514
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.904443825957881e-05,
-      "loss": 0.2424,
+      "loss": 0.1949,
       "step": 515
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9042582799888673e-05,
-      "loss": 0.1912,
+      "loss": 0.0864,
       "step": 516
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9040727340198535e-05,
-      "loss": 0.0512,
+      "loss": 0.1391,
       "step": 517
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.90388718805084e-05,
-      "loss": 0.2811,
+      "loss": 0.0867,
       "step": 518
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.9037016420818258e-05,
-      "loss": 0.05,
+      "loss": 0.0376,
       "step": 519
     },
     {
       "epoch": 0.14,
       "learning_rate": 1.903516096112812e-05,
-      "loss": 0.138,
+      "loss": 0.1387,
       "step": 520
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9033305501437982e-05,
-      "loss": 0.1485,
+      "loss": 0.1397,
       "step": 521
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9031450041747844e-05,
-      "loss": 0.1406,
+      "loss": 0.0883,
       "step": 522
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9029594582057706e-05,
-      "loss": 0.2298,
+      "loss": 0.137,
       "step": 523
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9027739122367568e-05,
-      "loss": 0.1478,
+      "loss": 0.0871,
       "step": 524
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.902588366267743e-05,
-      "loss": 0.1414,
+      "loss": 0.1385,
       "step": 525
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.902402820298729e-05,
-      "loss": 0.0929,
+      "loss": 0.1343,
       "step": 526
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9022172743297153e-05,
-      "loss": 0.2376,
+      "loss": 0.1377,
       "step": 527
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9020317283607015e-05,
-      "loss": 0.095,
+      "loss": 0.1378,
       "step": 528
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9018461823916877e-05,
-      "loss": 0.1909,
+      "loss": 0.0276,
       "step": 529
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.901660636422674e-05,
-      "loss": 0.1383,
+      "loss": 0.0263,
       "step": 530
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.90147509045366e-05,
-      "loss": 0.137,
+      "loss": 0.1359,
       "step": 531
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9012895444846463e-05,
-      "loss": 0.0931,
+      "loss": 0.259,
       "step": 532
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9011039985156324e-05,
-      "loss": 0.1443,
+      "loss": 0.1389,
       "step": 533
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9009184525466186e-05,
-      "loss": 0.1885,
+      "loss": 0.0814,
       "step": 534
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9007329065776048e-05,
-      "loss": 0.2411,
+      "loss": 0.0826,
       "step": 535
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.900547360608591e-05,
-      "loss": 0.1363,
+      "loss": 0.1948,
       "step": 536
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9003618146395772e-05,
-      "loss": 0.1374,
+      "loss": 0.0237,
       "step": 537
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.9001762686705634e-05,
-      "loss": 0.0894,
+      "loss": 0.0247,
       "step": 538
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.8999907227015496e-05,
-      "loss": 0.2485,
+      "loss": 0.2007,
       "step": 539
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.8998051767325358e-05,
-      "loss": 0.2366,
+      "loss": 0.1418,
       "step": 540
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.8996196307635216e-05,
-      "loss": 0.0362,
+      "loss": 0.2,
       "step": 541
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.899434084794508e-05,
-      "loss": 0.1389,
+      "loss": 0.1979,
       "step": 542
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.8992485388254943e-05,
-      "loss": 0.1387,
+      "loss": 0.0262,
       "step": 543
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.89906299285648e-05,
-      "loss": 0.0839,
+      "loss": 0.2027,
       "step": 544
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.8988774468874663e-05,
-      "loss": 0.0841,
+      "loss": 0.0814,
       "step": 545
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.898691900918453e-05,
-      "loss": 0.0884,
+      "loss": 0.0272,
       "step": 546
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.898506354949439e-05,
-      "loss": 0.0869,
+      "loss": 0.1384,
       "step": 547
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.898320808980425e-05,
-      "loss": 0.1883,
+      "loss": 0.0827,
       "step": 548
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.898135263011411e-05,
-      "loss": 0.1873,
+      "loss": 0.1365,
       "step": 549
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.8979497170423976e-05,
-      "loss": 0.1883,
+      "loss": 0.1392,
       "step": 550
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.8977641710733835e-05,
-      "loss": 0.2459,
+      "loss": 0.256,
       "step": 551
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.8975786251043697e-05,
-      "loss": 0.08,
+      "loss": 0.1384,
       "step": 552
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.897393079135356e-05,
-      "loss": 0.1895,
+      "loss": 0.0277,
       "step": 553
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.8972075331663424e-05,
-      "loss": 0.0824,
+      "loss": 0.1425,
       "step": 554
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.8970219871973282e-05,
-      "loss": 0.1402,
+      "loss": 0.1977,
       "step": 555
     },
     {
       "epoch": 0.15,
       "learning_rate": 1.8968364412283144e-05,
-      "loss": 0.0874,
+      "loss": 0.0292,
       "step": 556
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8966508952593006e-05,
-      "loss": 0.2468,
+      "loss": 0.2451,
       "step": 557
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8964653492902868e-05,
-      "loss": 0.085,
+      "loss": 0.1404,
       "step": 558
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.896279803321273e-05,
-      "loss": 0.1996,
+      "loss": 0.0292,
       "step": 559
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.896094257352259e-05,
-      "loss": 0.2485,
+      "loss": 0.2398,
       "step": 560
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8959087113832453e-05,
-      "loss": 0.0838,
+      "loss": 0.0825,
       "step": 561
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8957231654142315e-05,
-      "loss": 0.3482,
+      "loss": 0.03,
       "step": 562
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8955376194452177e-05,
-      "loss": 0.1365,
+      "loss": 0.085,
       "step": 563
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.895352073476204e-05,
-      "loss": 0.2969,
+      "loss": 0.0826,
       "step": 564
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.89516652750719e-05,
-      "loss": 0.1347,
+      "loss": 0.1971,
       "step": 565
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8949809815381763e-05,
-      "loss": 0.0851,
+      "loss": 0.0841,
       "step": 566
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8947954355691625e-05,
-      "loss": 0.1892,
+      "loss": 0.083,
       "step": 567
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8946098896001487e-05,
-      "loss": 0.0845,
+      "loss": 0.0254,
       "step": 568
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.894424343631135e-05,
-      "loss": 0.0895,
+      "loss": 0.0806,
       "step": 569
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.894238797662121e-05,
-      "loss": 0.0855,
+      "loss": 0.0804,
       "step": 570
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8940532516931072e-05,
-      "loss": 0.1932,
+      "loss": 0.2042,
       "step": 571
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8938677057240934e-05,
-      "loss": 0.2414,
+      "loss": 0.0798,
       "step": 572
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8936821597550792e-05,
-      "loss": 0.2423,
+      "loss": 0.0192,
       "step": 573
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8934966137860658e-05,
-      "loss": 0.1898,
+      "loss": 0.2004,
       "step": 574
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.893311067817052e-05,
-      "loss": 0.1406,
+      "loss": 0.2693,
       "step": 575
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.893125521848038e-05,
-      "loss": 0.1454,
+      "loss": 0.1449,
       "step": 576
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.892939975879024e-05,
-      "loss": 0.1882,
+      "loss": 0.0813,
       "step": 577
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8927544299100105e-05,
-      "loss": 0.0923,
+      "loss": 0.0223,
       "step": 578
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8925688839409967e-05,
-      "loss": 0.0929,
+      "loss": 0.2012,
       "step": 579
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8923833379719826e-05,
-      "loss": 0.1428,
+      "loss": 0.0795,
       "step": 580
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8921977920029687e-05,
-      "loss": 0.3895,
+      "loss": 0.1391,
       "step": 581
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8920122460339553e-05,
-      "loss": 0.0912,
+      "loss": 0.1409,
       "step": 582
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8918267000649415e-05,
-      "loss": 0.0925,
+      "loss": 0.2498,
       "step": 583
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8916411540959273e-05,
-      "loss": 0.0914,
+      "loss": 0.0856,
       "step": 584
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8914556081269135e-05,
-      "loss": 0.0926,
+      "loss": 0.194,
       "step": 585
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8912700621579e-05,
-      "loss": 0.1866,
+      "loss": 0.0325,
       "step": 586
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.891084516188886e-05,
-      "loss": 0.0869,
+      "loss": 0.0853,
       "step": 587
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.890898970219872e-05,
-      "loss": 0.1408,
+      "loss": 0.0331,
       "step": 588
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8907134242508582e-05,
-      "loss": 0.1398,
+      "loss": 0.1858,
       "step": 589
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8905278782818444e-05,
-      "loss": 0.1369,
+      "loss": 0.0814,
       "step": 590
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8903423323128306e-05,
-      "loss": 0.136,
+      "loss": 0.0297,
       "step": 591
     },
     {
       "epoch": 0.16,
       "learning_rate": 1.8901567863438168e-05,
-      "loss": 0.0862,
+      "loss": 0.0276,
       "step": 592
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.889971240374803e-05,
-      "loss": 0.2443,
+      "loss": 0.1453,
       "step": 593
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8897856944057892e-05,
-      "loss": 0.2474,
+      "loss": 0.0789,
       "step": 594
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8896001484367754e-05,
-      "loss": 0.0324,
+      "loss": 0.1393,
       "step": 595
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8894146024677616e-05,
-      "loss": 0.2429,
+      "loss": 0.0808,
       "step": 596
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8892290564987477e-05,
-      "loss": 0.0875,
+      "loss": 0.0845,
       "step": 597
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.889043510529734e-05,
-      "loss": 0.1416,
+      "loss": 0.0814,
       "step": 598
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.88885796456072e-05,
-      "loss": 0.1871,
+      "loss": 0.1403,
       "step": 599
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8886724185917063e-05,
-      "loss": 0.2495,
+      "loss": 0.1432,
       "step": 600
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8884868726226925e-05,
-      "loss": 0.1917,
+      "loss": 0.0198,
       "step": 601
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8883013266536787e-05,
-      "loss": 0.1424,
+      "loss": 0.0198,
       "step": 602
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.888115780684665e-05,
-      "loss": 0.0346,
+      "loss": 0.2102,
       "step": 603
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.887930234715651e-05,
-      "loss": 0.0839,
+      "loss": 0.0193,
       "step": 604
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8877446887466372e-05,
-      "loss": 0.1931,
+      "loss": 0.0803,
       "step": 605
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8875591427776234e-05,
-      "loss": 0.0875,
+      "loss": 0.0181,
       "step": 606
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8873735968086096e-05,
-      "loss": 0.0863,
+      "loss": 0.1419,
       "step": 607
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8871880508395958e-05,
-      "loss": 0.1763,
+      "loss": 0.0768,
       "step": 608
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8870025048705816e-05,
-      "loss": 0.1408,
+      "loss": 0.2057,
       "step": 609
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.886816958901568e-05,
-      "loss": 0.1382,
+      "loss": 0.2009,
       "step": 610
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8866314129325544e-05,
-      "loss": 0.1361,
+      "loss": 0.1325,
       "step": 611
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8864458669635402e-05,
-      "loss": 0.2503,
+      "loss": 0.1417,
       "step": 612
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8862603209945264e-05,
-      "loss": 0.1973,
+      "loss": 0.0783,
       "step": 613
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8860747750255126e-05,
-      "loss": 0.2489,
+      "loss": 0.0223,
       "step": 614
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.885889229056499e-05,
-      "loss": 0.0336,
+      "loss": 0.0819,
       "step": 615
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.885703683087485e-05,
-      "loss": 0.3505,
+      "loss": 0.0816,
       "step": 616
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.885518137118471e-05,
-      "loss": 0.1392,
+      "loss": 0.1424,
       "step": 617
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8853325911494573e-05,
-      "loss": 0.1397,
+      "loss": 0.0835,
       "step": 618
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8851470451804435e-05,
-      "loss": 0.1379,
+      "loss": 0.0788,
       "step": 619
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8849614992114297e-05,
-      "loss": 0.1439,
+      "loss": 0.1939,
       "step": 620
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.884775953242416e-05,
-      "loss": 0.1388,
+      "loss": 0.143,
       "step": 621
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.884590407273402e-05,
-      "loss": 0.1465,
+      "loss": 0.1397,
       "step": 622
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8844048613043883e-05,
-      "loss": 0.0404,
+      "loss": 0.1427,
       "step": 623
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8842193153353744e-05,
-      "loss": 0.1895,
+      "loss": 0.0794,
       "step": 624
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8840337693663606e-05,
-      "loss": 0.245,
+      "loss": 0.0262,
       "step": 625
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8838482233973468e-05,
-      "loss": 0.0392,
+      "loss": 0.1336,
       "step": 626
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.883662677428333e-05,
-      "loss": 0.1939,
+      "loss": 0.1968,
       "step": 627
     },
     {
       "epoch": 0.17,
       "learning_rate": 1.8834771314593192e-05,
-      "loss": 0.2942,
+      "loss": 0.086,
       "step": 628
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8832915854903054e-05,
-      "loss": 0.0906,
+      "loss": 0.144,
       "step": 629
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8831060395212916e-05,
-      "loss": 0.1967,
+      "loss": 0.1914,
       "step": 630
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8829204935522778e-05,
-      "loss": 0.1432,
+      "loss": 0.0839,
       "step": 631
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.882734947583264e-05,
-      "loss": 0.1455,
+      "loss": 0.0293,
       "step": 632
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.88254940161425e-05,
-      "loss": 0.1409,
+      "loss": 0.0274,
       "step": 633
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8823638556452363e-05,
-      "loss": 0.0854,
+      "loss": 0.0861,
       "step": 634
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8821783096762225e-05,
-      "loss": 0.1416,
+      "loss": 0.1993,
       "step": 635
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8819927637072087e-05,
-      "loss": 0.0892,
+      "loss": 0.139,
       "step": 636
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.881807217738195e-05,
-      "loss": 0.0366,
+      "loss": 0.0239,
       "step": 637
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8816216717691807e-05,
-      "loss": 0.0328,
+      "loss": 0.086,
       "step": 638
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8814361258001673e-05,
-      "loss": 0.0833,
+      "loss": 0.2093,
       "step": 639
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8812505798311534e-05,
-      "loss": 0.1433,
+      "loss": 0.1366,
       "step": 640
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8810650338621393e-05,
-      "loss": 0.1939,
+      "loss": 0.1368,
       "step": 641
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8808794878931255e-05,
-      "loss": 0.1443,
+      "loss": 0.08,
       "step": 642
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.880693941924112e-05,
-      "loss": 0.2485,
+      "loss": 0.1412,
       "step": 643
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8805083959550982e-05,
-      "loss": 0.2017,
+      "loss": 0.0822,
       "step": 644
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.880322849986084e-05,
-      "loss": 0.3584,
+      "loss": 0.0784,
       "step": 645
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8801373040170702e-05,
-      "loss": 0.0806,
+      "loss": 0.1446,
       "step": 646
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8799517580480568e-05,
-      "loss": 0.1373,
+      "loss": 0.0808,
       "step": 647
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8797662120790426e-05,
-      "loss": 0.0848,
+      "loss": 0.1406,
       "step": 648
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8795806661100288e-05,
-      "loss": 0.4113,
+      "loss": 0.0791,
       "step": 649
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.879395120141015e-05,
-      "loss": 0.1426,
+      "loss": 0.0838,
       "step": 650
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8792095741720015e-05,
-      "loss": 0.0879,
+      "loss": 0.1981,
       "step": 651
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8790240282029873e-05,
-      "loss": 0.142,
+      "loss": 0.0814,
       "step": 652
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8788384822339735e-05,
-      "loss": 0.0343,
+      "loss": 0.0824,
       "step": 653
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8786529362649597e-05,
-      "loss": 0.1895,
+      "loss": 0.14,
       "step": 654
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.878467390295946e-05,
-      "loss": 0.1398,
+      "loss": 0.0231,
       "step": 655
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.878281844326932e-05,
-      "loss": 0.194,
+      "loss": 0.0233,
       "step": 656
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8780962983579183e-05,
-      "loss": 0.0892,
+      "loss": 0.1392,
       "step": 657
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8779107523889045e-05,
-      "loss": 0.136,
+      "loss": 0.0187,
       "step": 658
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8777252064198907e-05,
-      "loss": 0.1929,
+      "loss": 0.1393,
       "step": 659
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.877539660450877e-05,
-      "loss": 0.1401,
+      "loss": 0.0832,
       "step": 660
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.877354114481863e-05,
-      "loss": 0.1914,
+      "loss": 0.0783,
       "step": 661
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8771685685128492e-05,
-      "loss": 0.1414,
+      "loss": 0.08,
       "step": 662
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8769830225438354e-05,
-      "loss": 0.1396,
+      "loss": 0.2705,
       "step": 663
     },
     {
       "epoch": 0.18,
       "learning_rate": 1.8767974765748216e-05,
-      "loss": 0.09,
+      "loss": 0.2067,
       "step": 664
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8766119306058078e-05,
-      "loss": 0.086,
+      "loss": 0.084,
       "step": 665
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.876426384636794e-05,
-      "loss": 0.1386,
+      "loss": 0.0787,
       "step": 666
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.87624083866778e-05,
-      "loss": 0.1911,
+      "loss": 0.0858,
       "step": 667
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8760552926987663e-05,
-      "loss": 0.2376,
+      "loss": 0.1871,
       "step": 668
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8758697467297525e-05,
-      "loss": 0.187,
+      "loss": 0.035,
       "step": 669
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8756842007607384e-05,
-      "loss": 0.1416,
+      "loss": 0.0838,
       "step": 670
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.875498654791725e-05,
-      "loss": 0.3411,
+      "loss": 0.0855,
       "step": 671
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.875313108822711e-05,
-      "loss": 0.0852,
+      "loss": 0.1364,
       "step": 672
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8751275628536973e-05,
-      "loss": 0.1876,
+      "loss": 0.1967,
       "step": 673
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.874942016884683e-05,
-      "loss": 0.1857,
+      "loss": 0.0295,
       "step": 674
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8747564709156697e-05,
-      "loss": 0.1382,
+      "loss": 0.1383,
       "step": 675
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.874570924946656e-05,
-      "loss": 0.1897,
+      "loss": 0.0876,
       "step": 676
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8743853789776417e-05,
-      "loss": 0.1436,
+      "loss": 0.0267,
       "step": 677
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.874199833008628e-05,
-      "loss": 0.1436,
+      "loss": 0.0814,
       "step": 678
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8740142870396144e-05,
-      "loss": 0.1483,
+      "loss": 0.0815,
       "step": 679
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8738287410706006e-05,
-      "loss": 0.0463,
+      "loss": 0.2013,
       "step": 680
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8736431951015864e-05,
-      "loss": 0.2409,
+      "loss": 0.0828,
       "step": 681
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8734576491325726e-05,
-      "loss": 0.1879,
+      "loss": 0.0828,
       "step": 682
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.873272103163559e-05,
-      "loss": 0.0922,
+      "loss": 0.1432,
       "step": 683
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.873086557194545e-05,
-      "loss": 0.2384,
+      "loss": 0.0212,
       "step": 684
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8729010112255312e-05,
-      "loss": 0.0361,
+      "loss": 0.2667,
       "step": 685
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8727154652565174e-05,
-      "loss": 0.25,
+      "loss": 0.0204,
       "step": 686
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8725299192875036e-05,
-      "loss": 0.0845,
+      "loss": 0.2642,
       "step": 687
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8723443733184897e-05,
-      "loss": 0.1385,
+      "loss": 0.0834,
       "step": 688
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.872158827349476e-05,
-      "loss": 0.1349,
+      "loss": 0.0827,
       "step": 689
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.871973281380462e-05,
-      "loss": 0.1401,
+      "loss": 0.1416,
       "step": 690
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8717877354114483e-05,
-      "loss": 0.1384,
+      "loss": 0.0823,
       "step": 691
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8716021894424345e-05,
-      "loss": 0.0873,
+      "loss": 0.1378,
       "step": 692
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8714166434734207e-05,
-      "loss": 0.1911,
+      "loss": 0.0869,
       "step": 693
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.871231097504407e-05,
-      "loss": 0.1955,
+      "loss": 0.1383,
       "step": 694
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.871045551535393e-05,
-      "loss": 0.137,
+      "loss": 0.0332,
       "step": 695
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8708600055663792e-05,
-      "loss": 0.1353,
+      "loss": 0.0826,
       "step": 696
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8706744595973654e-05,
-      "loss": 0.1982,
+      "loss": 0.1411,
       "step": 697
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8704889136283516e-05,
-      "loss": 0.1922,
+      "loss": 0.1404,
       "step": 698
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.8703033676593378e-05,
-      "loss": 0.1961,
+      "loss": 0.0219,
       "step": 699
     },
     {
       "epoch": 0.19,
       "learning_rate": 1.870117821690324e-05,
-      "loss": 0.1886,
+      "loss": 0.1467,
       "step": 700
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8699322757213102e-05,
-      "loss": 0.3536,
+      "loss": 0.0208,
       "step": 701
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8697467297522964e-05,
-      "loss": 0.0868,
+      "loss": 0.0199,
       "step": 702
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8695611837832825e-05,
-      "loss": 0.1397,
+      "loss": 0.3242,
       "step": 703
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8693756378142687e-05,
-      "loss": 0.1369,
+      "loss": 0.0196,
       "step": 704
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.869190091845255e-05,
-      "loss": 0.0894,
+      "loss": 0.081,
       "step": 705
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8690045458762408e-05,
-      "loss": 0.2964,
+      "loss": 0.1493,
       "step": 706
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8688189999072273e-05,
-      "loss": 0.0872,
+      "loss": 0.0766,
       "step": 707
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8686334539382135e-05,
-      "loss": 0.1473,
+      "loss": 0.0839,
       "step": 708
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8684479079691997e-05,
-      "loss": 0.0406,
+      "loss": 0.0811,
       "step": 709
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8682623620001855e-05,
-      "loss": 0.293,
+      "loss": 0.2627,
       "step": 710
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.868076816031172e-05,
-      "loss": 0.1408,
+      "loss": 0.3265,
       "step": 711
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8678912700621582e-05,
-      "loss": 0.2396,
+      "loss": 0.0812,
       "step": 712
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.867705724093144e-05,
-      "loss": 0.1943,
+      "loss": 0.2646,
       "step": 713
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8675201781241303e-05,
-      "loss": 0.1915,
+      "loss": 0.1397,
       "step": 714
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8673346321551168e-05,
-      "loss": 0.1465,
+      "loss": 0.1947,
       "step": 715
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8671490861861026e-05,
-      "loss": 0.0945,
+      "loss": 0.1951,
       "step": 716
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8669635402170888e-05,
-      "loss": 0.432,
+      "loss": 0.0284,
       "step": 717
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.866777994248075e-05,
-      "loss": 0.0898,
+      "loss": 0.0293,
       "step": 718
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8665924482790615e-05,
-      "loss": 0.0428,
+      "loss": 0.086,
       "step": 719
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8664069023100474e-05,
-      "loss": 0.1935,
+      "loss": 0.2515,
       "step": 720
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8662213563410336e-05,
-      "loss": 0.1906,
+      "loss": 0.0814,
       "step": 721
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8660358103720198e-05,
-      "loss": 0.0911,
+      "loss": 0.1879,
       "step": 722
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.865850264403006e-05,
-      "loss": 0.0427,
+      "loss": 0.1409,
       "step": 723
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.865664718433992e-05,
-      "loss": 0.0394,
+      "loss": 0.1314,
       "step": 724
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8654791724649783e-05,
-      "loss": 0.1369,
+      "loss": 0.1425,
       "step": 725
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8652936264959645e-05,
-      "loss": 0.089,
+      "loss": 0.1905,
       "step": 726
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8651080805269507e-05,
-      "loss": 0.2476,
+      "loss": 0.1371,
       "step": 727
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.864922534557937e-05,
-      "loss": 0.1865,
+      "loss": 0.1382,
       "step": 728
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.864736988588923e-05,
-      "loss": 0.0903,
+      "loss": 0.182,
       "step": 729
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8645514426199093e-05,
-      "loss": 0.0314,
+      "loss": 0.1915,
       "step": 730
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8643658966508954e-05,
-      "loss": 0.0309,
+      "loss": 0.1418,
       "step": 731
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8641803506818816e-05,
-      "loss": 0.1417,
+      "loss": 0.186,
       "step": 732
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8639948047128678e-05,
-      "loss": 0.0878,
+      "loss": 0.0428,
       "step": 733
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.863809258743854e-05,
-      "loss": 0.0856,
+      "loss": 0.0897,
       "step": 734
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8636237127748402e-05,
-      "loss": 0.138,
+      "loss": 0.0415,
       "step": 735
     },
     {
       "epoch": 0.2,
       "learning_rate": 1.8634381668058264e-05,
-      "loss": 0.1359,
+      "loss": 0.1873,
       "step": 736
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8632526208368126e-05,
-      "loss": 0.199,
+      "loss": 0.1444,
       "step": 737
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8630670748677984e-05,
-      "loss": 0.2567,
+      "loss": 0.0342,
       "step": 738
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.862881528898785e-05,
-      "loss": 0.26,
+      "loss": 0.0302,
       "step": 739
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.862695982929771e-05,
-      "loss": 0.2569,
+      "loss": 0.0843,
       "step": 740
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8625104369607573e-05,
-      "loss": 0.1975,
+      "loss": 0.0254,
       "step": 741
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.862324890991743e-05,
-      "loss": 0.0266,
+      "loss": 0.1382,
       "step": 742
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8621393450227297e-05,
-      "loss": 0.0785,
+      "loss": 0.2597,
       "step": 743
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.861953799053716e-05,
-      "loss": 0.2531,
+      "loss": 0.3214,
       "step": 744
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8617682530847017e-05,
-      "loss": 0.2506,
+      "loss": 0.1972,
       "step": 745
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.861582707115688e-05,
-      "loss": 0.1417,
+      "loss": 0.0208,
       "step": 746
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8613971611466744e-05,
-      "loss": 0.0304,
+      "loss": 0.197,
       "step": 747
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8612116151776606e-05,
-      "loss": 0.303,
+      "loss": 0.2009,
       "step": 748
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8610260692086465e-05,
-      "loss": 0.1356,
+      "loss": 0.1421,
       "step": 749
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8608405232396327e-05,
-      "loss": 0.0819,
+      "loss": 0.078,
       "step": 750
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8606549772706192e-05,
-      "loss": 0.2461,
+      "loss": 0.0817,
       "step": 751
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.860469431301605e-05,
-      "loss": 0.0344,
+      "loss": 0.3113,
       "step": 752
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8602838853325912e-05,
-      "loss": 0.1405,
+      "loss": 0.1415,
       "step": 753
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8600983393635774e-05,
-      "loss": 0.1382,
+      "loss": 0.2554,
       "step": 754
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8599127933945636e-05,
-      "loss": 0.1903,
+      "loss": 0.0295,
       "step": 755
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8597272474255498e-05,
-      "loss": 0.2482,
+      "loss": 0.0872,
       "step": 756
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.859541701456536e-05,
-      "loss": 0.1392,
+      "loss": 0.2471,
       "step": 757
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.859356155487522e-05,
-      "loss": 0.2406,
+      "loss": 0.242,
       "step": 758
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8591706095185083e-05,
-      "loss": 0.0889,
+      "loss": 0.0894,
       "step": 759
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8589850635494945e-05,
-      "loss": 0.2868,
+      "loss": 0.0909,
       "step": 760
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8587995175804807e-05,
-      "loss": 0.0921,
+      "loss": 0.1355,
       "step": 761
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.858613971611467e-05,
-      "loss": 0.0409,
+      "loss": 0.1941,
       "step": 762
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.858428425642453e-05,
-      "loss": 0.1915,
+      "loss": 0.091,
       "step": 763
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8582428796734393e-05,
-      "loss": 0.2436,
+      "loss": 0.0904,
       "step": 764
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8580573337044255e-05,
-      "loss": 0.141,
+      "loss": 0.0822,
       "step": 765
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8578717877354117e-05,
-      "loss": 0.2465,
+      "loss": 0.0404,
       "step": 766
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8576862417663975e-05,
-      "loss": 0.2915,
+      "loss": 0.089,
       "step": 767
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.857500695797384e-05,
-      "loss": 0.0451,
+      "loss": 0.1391,
       "step": 768
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8573151498283702e-05,
-      "loss": 0.1407,
+      "loss": 0.0333,
       "step": 769
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8571296038593564e-05,
-      "loss": 0.0924,
+      "loss": 0.139,
       "step": 770
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8569440578903423e-05,
-      "loss": 0.146,
+      "loss": 0.0824,
       "step": 771
     },
     {
       "epoch": 0.21,
       "learning_rate": 1.8567585119213288e-05,
-      "loss": 0.2357,
+      "loss": 0.0821,
       "step": 772
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.856572965952315e-05,
-      "loss": 0.0432,
+      "loss": 0.0784,
       "step": 773
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8563874199833008e-05,
-      "loss": 0.0405,
+      "loss": 0.0804,
       "step": 774
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.856201874014287e-05,
-      "loss": 0.1433,
+      "loss": 0.1405,
       "step": 775
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8560163280452735e-05,
-      "loss": 0.1868,
+      "loss": 0.198,
       "step": 776
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8558307820762597e-05,
-      "loss": 0.0381,
+      "loss": 0.0192,
       "step": 777
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8556452361072456e-05,
-      "loss": 0.1903,
+      "loss": 0.0836,
       "step": 778
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8554596901382317e-05,
-      "loss": 0.0856,
+      "loss": 0.081,
       "step": 779
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8552741441692183e-05,
-      "loss": 0.0862,
+      "loss": 0.0162,
       "step": 780
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.855088598200204e-05,
-      "loss": 0.0893,
+      "loss": 0.1432,
       "step": 781
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8549030522311903e-05,
-      "loss": 0.1415,
+      "loss": 0.1492,
       "step": 782
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8547175062621765e-05,
-      "loss": 0.1946,
+      "loss": 0.1469,
       "step": 783
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8545319602931627e-05,
-      "loss": 0.1444,
+      "loss": 0.0841,
       "step": 784
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.854346414324149e-05,
-      "loss": 0.189,
+      "loss": 0.016,
       "step": 785
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.854160868355135e-05,
-      "loss": 0.1425,
+      "loss": 0.1436,
       "step": 786
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8539753223861212e-05,
-      "loss": 0.2529,
+      "loss": 0.078,
       "step": 787
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8537897764171074e-05,
-      "loss": 0.0835,
+      "loss": 0.2758,
       "step": 788
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8536042304480936e-05,
-      "loss": 0.1891,
+      "loss": 0.079,
       "step": 789
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8534186844790798e-05,
-      "loss": 0.1903,
+      "loss": 0.325,
       "step": 790
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.853233138510066e-05,
-      "loss": 0.2476,
+      "loss": 0.0209,
       "step": 791
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8530475925410522e-05,
-      "loss": 0.2527,
+      "loss": 0.1376,
       "step": 792
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8528620465720384e-05,
-      "loss": 0.3016,
+      "loss": 0.1378,
       "step": 793
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8526765006030246e-05,
-      "loss": 0.0331,
+      "loss": 0.1911,
       "step": 794
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8524909546340107e-05,
-      "loss": 0.1358,
+      "loss": 0.0811,
       "step": 795
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.852305408664997e-05,
-      "loss": 0.1878,
+      "loss": 0.0807,
       "step": 796
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.852119862695983e-05,
-      "loss": 0.1373,
+      "loss": 0.1365,
       "step": 797
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8519343167269693e-05,
-      "loss": 0.1362,
+      "loss": 0.0828,
       "step": 798
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8517487707579555e-05,
-      "loss": 0.1388,
+      "loss": 0.1358,
       "step": 799
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8515632247889417e-05,
-      "loss": 0.3926,
+      "loss": 0.0821,
       "step": 800
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.851377678819928e-05,
-      "loss": 0.0892,
+      "loss": 0.1919,
       "step": 801
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.851192132850914e-05,
-      "loss": 0.038,
+      "loss": 0.0308,
       "step": 802
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8510065868819e-05,
-      "loss": 0.1363,
+      "loss": 0.1338,
       "step": 803
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8508210409128864e-05,
-      "loss": 0.0868,
+      "loss": 0.2985,
       "step": 804
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8506354949438726e-05,
-      "loss": 0.1916,
+      "loss": 0.1393,
       "step": 805
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8504499489748588e-05,
-      "loss": 0.0409,
+      "loss": 0.1392,
       "step": 806
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8502644030058446e-05,
-      "loss": 0.0897,
+      "loss": 0.0843,
       "step": 807
     },
     {
       "epoch": 0.22,
       "learning_rate": 1.8500788570368312e-05,
-      "loss": 0.1381,
+      "loss": 0.1421,
       "step": 808
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8498933110678174e-05,
-      "loss": 0.1902,
+      "loss": 0.1428,
       "step": 809
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8497077650988032e-05,
-      "loss": 0.14,
+      "loss": 0.0882,
       "step": 810
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8495222191297894e-05,
-      "loss": 0.1945,
+      "loss": 0.1366,
       "step": 811
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.849336673160776e-05,
-      "loss": 0.1384,
+      "loss": 0.0328,
       "step": 812
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8491511271917618e-05,
-      "loss": 0.1394,
+      "loss": 0.0859,
       "step": 813
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.848965581222748e-05,
-      "loss": 0.1906,
+      "loss": 0.0825,
       "step": 814
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.848780035253734e-05,
-      "loss": 0.0356,
+      "loss": 0.084,
       "step": 815
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8485944892847207e-05,
-      "loss": 0.0866,
+      "loss": 0.1358,
       "step": 816
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8484089433157065e-05,
-      "loss": 0.142,
+      "loss": 0.1411,
       "step": 817
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8482233973466927e-05,
-      "loss": 0.1389,
+      "loss": 0.0248,
       "step": 818
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.848037851377679e-05,
-      "loss": 0.1972,
+      "loss": 0.0252,
       "step": 819
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.847852305408665e-05,
-      "loss": 0.2426,
+      "loss": 0.1963,
       "step": 820
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8476667594396513e-05,
-      "loss": 0.1371,
+      "loss": 0.1912,
       "step": 821
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8474812134706375e-05,
-      "loss": 0.1386,
+      "loss": 0.1442,
       "step": 822
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8472956675016236e-05,
-      "loss": 0.2433,
+      "loss": 0.1422,
       "step": 823
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8471101215326098e-05,
-      "loss": 0.0328,
+      "loss": 0.0821,
       "step": 824
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.846924575563596e-05,
-      "loss": 0.1913,
+      "loss": 0.0204,
       "step": 825
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8467390295945822e-05,
-      "loss": 0.2465,
+      "loss": 0.079,
       "step": 826
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8465534836255684e-05,
-      "loss": 0.0853,
+      "loss": 0.0199,
       "step": 827
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8463679376565546e-05,
-      "loss": 0.0879,
+      "loss": 0.1435,
       "step": 828
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8461823916875408e-05,
-      "loss": 0.2489,
+      "loss": 0.0173,
       "step": 829
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.845996845718527e-05,
-      "loss": 0.2957,
+      "loss": 0.2703,
       "step": 830
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.845811299749513e-05,
-      "loss": 0.1924,
+      "loss": 0.2066,
       "step": 831
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8456257537804993e-05,
-      "loss": 0.0833,
+      "loss": 0.0189,
       "step": 832
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8454402078114855e-05,
-      "loss": 0.3483,
+      "loss": 0.1413,
       "step": 833
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8452546618424717e-05,
-      "loss": 0.1378,
+      "loss": 0.0187,
       "step": 834
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.845069115873458e-05,
-      "loss": 0.2427,
+      "loss": 0.1334,
       "step": 835
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.844883569904444e-05,
-      "loss": 0.2362,
+      "loss": 0.0207,
       "step": 836
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8446980239354303e-05,
-      "loss": 0.0423,
+      "loss": 0.08,
       "step": 837
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8445124779664164e-05,
-      "loss": 0.0948,
+      "loss": 0.1339,
       "step": 838
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8443269319974023e-05,
-      "loss": 0.0909,
+      "loss": 0.0751,
       "step": 839
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8441413860283888e-05,
-      "loss": 0.489,
+      "loss": 0.2674,
       "step": 840
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.843955840059375e-05,
-      "loss": 0.1409,
+      "loss": 0.1931,
       "step": 841
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.843770294090361e-05,
-      "loss": 0.1425,
+      "loss": 0.0827,
       "step": 842
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.843584748121347e-05,
-      "loss": 0.2924,
+      "loss": 0.024,
       "step": 843
     },
     {
       "epoch": 0.23,
       "learning_rate": 1.8433992021523336e-05,
-      "loss": 0.2415,
+      "loss": 0.0815,
       "step": 844
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8432136561833198e-05,
-      "loss": 0.1375,
+      "loss": 0.082,
       "step": 845
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8430281102143056e-05,
-      "loss": 0.2825,
+      "loss": 0.0808,
       "step": 846
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8428425642452918e-05,
-      "loss": 0.1938,
+      "loss": 0.1374,
       "step": 847
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8426570182762783e-05,
-      "loss": 0.0965,
+      "loss": 0.0248,
       "step": 848
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.842471472307264e-05,
-      "loss": 0.1933,
+      "loss": 0.2411,
       "step": 849
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8422859263382504e-05,
-      "loss": 0.1385,
+      "loss": 0.2028,
       "step": 850
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8421003803692365e-05,
-      "loss": 0.2861,
+      "loss": 0.137,
       "step": 851
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8419148344002227e-05,
-      "loss": 0.0494,
+      "loss": 0.1371,
       "step": 852
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.841729288431209e-05,
-      "loss": 0.2347,
+      "loss": 0.0292,
       "step": 853
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.841543742462195e-05,
-      "loss": 0.14,
+      "loss": 0.0285,
       "step": 854
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8413581964931813e-05,
-      "loss": 0.3759,
+      "loss": 0.2391,
       "step": 855
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8411726505241675e-05,
-      "loss": 0.096,
+      "loss": 0.4086,
       "step": 856
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8409871045551537e-05,
-      "loss": 0.1878,
+      "loss": 0.0794,
       "step": 857
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.84080155858614e-05,
-      "loss": 0.1433,
+      "loss": 0.0305,
       "step": 858
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.840616012617126e-05,
-      "loss": 0.1443,
+      "loss": 0.0357,
       "step": 859
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8404304666481122e-05,
-      "loss": 0.0944,
+      "loss": 0.1862,
       "step": 860
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8402449206790984e-05,
-      "loss": 0.1437,
+      "loss": 0.1371,
       "step": 861
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8400593747100846e-05,
-      "loss": 0.1432,
+      "loss": 0.0875,
       "step": 862
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8398738287410708e-05,
-      "loss": 0.1936,
+      "loss": 0.0836,
       "step": 863
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.839688282772057e-05,
-      "loss": 0.0923,
+      "loss": 0.0789,
       "step": 864
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.839502736803043e-05,
-      "loss": 0.1886,
+      "loss": 0.2,
       "step": 865
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8393171908340293e-05,
-      "loss": 0.1934,
+      "loss": 0.0284,
       "step": 866
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8391316448650155e-05,
-      "loss": 0.1908,
+      "loss": 0.0822,
       "step": 867
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8389460988960017e-05,
-      "loss": 0.1917,
+      "loss": 0.1857,
       "step": 868
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.838760552926988e-05,
-      "loss": 0.1447,
+      "loss": 0.0805,
       "step": 869
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.838575006957974e-05,
-      "loss": 0.1428,
+      "loss": 0.1942,
       "step": 870
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.83838946098896e-05,
-      "loss": 0.3384,
+      "loss": 0.0806,
       "step": 871
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8382039150199465e-05,
-      "loss": 0.1922,
+      "loss": 0.1359,
       "step": 872
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8380183690509327e-05,
-      "loss": 0.1947,
+      "loss": 0.1381,
       "step": 873
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.837832823081919e-05,
-      "loss": 0.0913,
+      "loss": 0.0852,
       "step": 874
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8376472771129047e-05,
-      "loss": 0.088,
+      "loss": 0.0799,
       "step": 875
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8374617311438912e-05,
-      "loss": 0.1428,
+      "loss": 0.1909,
       "step": 876
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8372761851748774e-05,
-      "loss": 0.2466,
+      "loss": 0.0801,
       "step": 877
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8370906392058633e-05,
-      "loss": 0.2431,
+      "loss": 0.0776,
       "step": 878
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.8369050932368494e-05,
-      "loss": 0.1949,
+      "loss": 0.1338,
       "step": 879
     },
     {
       "epoch": 0.24,
       "learning_rate": 1.836719547267836e-05,
-      "loss": 0.091,
+      "loss": 0.0798,
       "step": 880
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8365340012988218e-05,
-      "loss": 0.1399,
+      "loss": 0.1352,
       "step": 881
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.836348455329808e-05,
-      "loss": 0.1432,
+      "loss": 0.073,
       "step": 882
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8361629093607942e-05,
-      "loss": 0.0402,
+      "loss": 0.1309,
       "step": 883
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8359773633917807e-05,
-      "loss": 0.192,
+      "loss": 0.1902,
       "step": 884
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8357918174227666e-05,
-      "loss": 0.1901,
+      "loss": 0.0218,
       "step": 885
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8356062714537527e-05,
-      "loss": 0.1403,
+      "loss": 0.1394,
       "step": 886
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.835420725484739e-05,
-      "loss": 0.1434,
+      "loss": 0.0795,
       "step": 887
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.835235179515725e-05,
-      "loss": 0.1412,
+      "loss": 0.0188,
       "step": 888
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8350496335467113e-05,
-      "loss": 0.1369,
+      "loss": 0.076,
       "step": 889
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8348640875776975e-05,
-      "loss": 0.1896,
+      "loss": 0.1501,
       "step": 890
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8346785416086837e-05,
-      "loss": 0.2451,
+      "loss": 0.0783,
       "step": 891
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.83449299563967e-05,
-      "loss": 0.138,
+      "loss": 0.1339,
       "step": 892
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.834307449670656e-05,
-      "loss": 0.1896,
+      "loss": 0.0788,
       "step": 893
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8341219037016422e-05,
-      "loss": 0.1409,
+      "loss": 0.0779,
       "step": 894
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8339363577326284e-05,
-      "loss": 0.0376,
+      "loss": 0.1405,
       "step": 895
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8337508117636146e-05,
-      "loss": 0.2419,
+      "loss": 0.08,
       "step": 896
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8335652657946008e-05,
-      "loss": 0.1412,
+      "loss": 0.0796,
       "step": 897
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.833379719825587e-05,
-      "loss": 0.1925,
+      "loss": 0.1926,
       "step": 898
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8331941738565732e-05,
-      "loss": 0.1394,
+      "loss": 0.1844,
       "step": 899
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8330086278875594e-05,
-      "loss": 0.1398,
+      "loss": 0.0841,
       "step": 900
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8328230819185456e-05,
-      "loss": 0.1395,
+      "loss": 0.0277,
       "step": 901
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8326375359495317e-05,
-      "loss": 0.1925,
+      "loss": 0.0791,
       "step": 902
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.832451989980518e-05,
-      "loss": 0.2827,
+      "loss": 0.1955,
       "step": 903
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.832266444011504e-05,
-      "loss": 0.2443,
+      "loss": 0.1369,
       "step": 904
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8320808980424903e-05,
-      "loss": 0.0887,
+      "loss": 0.1381,
       "step": 905
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8318953520734765e-05,
-      "loss": 0.1946,
+      "loss": 0.03,
       "step": 906
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8317098061044623e-05,
-      "loss": 0.2436,
+      "loss": 0.1846,
       "step": 907
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.831524260135449e-05,
-      "loss": 0.0916,
+      "loss": 0.0851,
       "step": 908
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.831338714166435e-05,
-      "loss": 0.0885,
+      "loss": 0.1351,
       "step": 909
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.831153168197421e-05,
-      "loss": 0.2862,
+      "loss": 0.0285,
       "step": 910
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.830967622228407e-05,
-      "loss": 0.1871,
+      "loss": 0.2467,
       "step": 911
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8307820762593933e-05,
-      "loss": 0.137,
+      "loss": 0.1333,
       "step": 912
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8305965302903798e-05,
-      "loss": 0.2377,
+      "loss": 0.0818,
       "step": 913
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.8304109843213656e-05,
-      "loss": 0.0431,
+      "loss": 0.1341,
       "step": 914
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.830225438352352e-05,
-      "loss": 0.0432,
+      "loss": 0.0869,
       "step": 915
     },
     {
       "epoch": 0.25,
       "learning_rate": 1.830039892383338e-05,
-      "loss": 0.2377,
+      "loss": 0.0252,
       "step": 916
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8298543464143242e-05,
-      "loss": 0.0914,
+      "loss": 0.205,
       "step": 917
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8296688004453104e-05,
-      "loss": 0.0884,
+      "loss": 0.0822,
       "step": 918
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8294832544762966e-05,
-      "loss": 0.2419,
+      "loss": 0.1833,
       "step": 919
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8292977085072828e-05,
-      "loss": 0.1946,
+      "loss": 0.1899,
       "step": 920
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.829112162538269e-05,
-      "loss": 0.0896,
+      "loss": 0.1942,
       "step": 921
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.828926616569255e-05,
-      "loss": 0.2902,
+      "loss": 0.0319,
       "step": 922
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8287410706002413e-05,
-      "loss": 0.2425,
+      "loss": 0.0334,
       "step": 923
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8285555246312275e-05,
-      "loss": 0.2425,
+      "loss": 0.081,
       "step": 924
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8283699786622137e-05,
-      "loss": 0.1819,
+      "loss": 0.0302,
       "step": 925
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8281844326932e-05,
-      "loss": 0.1408,
+      "loss": 0.2353,
       "step": 926
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.827998886724186e-05,
-      "loss": 0.1401,
+      "loss": 0.0255,
       "step": 927
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8278133407551723e-05,
-      "loss": 0.1358,
+      "loss": 0.0779,
       "step": 928
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8276277947861585e-05,
-      "loss": 0.1915,
+      "loss": 0.1893,
       "step": 929
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8274422488171446e-05,
-      "loss": 0.0416,
+      "loss": 0.0822,
       "step": 930
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8272567028481308e-05,
-      "loss": 0.1444,
+      "loss": 0.1322,
       "step": 931
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.827071156879117e-05,
-      "loss": 0.1403,
+      "loss": 0.0746,
       "step": 932
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8268856109101032e-05,
-      "loss": 0.2426,
+      "loss": 0.1905,
       "step": 933
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8267000649410894e-05,
-      "loss": 0.142,
+      "loss": 0.192,
       "step": 934
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8265145189720756e-05,
-      "loss": 0.2423,
+      "loss": 0.023,
       "step": 935
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8263289730030614e-05,
-      "loss": 0.142,
+      "loss": 0.1333,
       "step": 936
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.826143427034048e-05,
-      "loss": 0.2408,
+      "loss": 0.0796,
       "step": 937
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.825957881065034e-05,
-      "loss": 0.1389,
+      "loss": 0.023,
       "step": 938
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.82577233509602e-05,
-      "loss": 0.1324,
+      "loss": 0.1857,
       "step": 939
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8255867891270062e-05,
-      "loss": 0.0878,
+      "loss": 0.0206,
       "step": 940
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8254012431579927e-05,
-      "loss": 0.1356,
+      "loss": 0.1932,
       "step": 941
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.825215697188979e-05,
-      "loss": 0.0403,
+      "loss": 0.0724,
       "step": 942
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8250301512199647e-05,
-      "loss": 0.1413,
+      "loss": 0.0783,
       "step": 943
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.824844605250951e-05,
-      "loss": 0.2335,
+      "loss": 0.1351,
       "step": 944
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8246590592819374e-05,
-      "loss": 0.1869,
+      "loss": 0.0802,
       "step": 945
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8244735133129233e-05,
-      "loss": 0.0885,
+      "loss": 0.2364,
       "step": 946
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8242879673439095e-05,
-      "loss": 0.1915,
+      "loss": 0.0846,
       "step": 947
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8241024213748957e-05,
-      "loss": 0.1364,
+      "loss": 0.0847,
       "step": 948
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8239168754058822e-05,
-      "loss": 0.1402,
+      "loss": 0.0845,
       "step": 949
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.823731329436868e-05,
-      "loss": 0.1951,
+      "loss": 0.0783,
       "step": 950
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8235457834678542e-05,
-      "loss": 0.3039,
+      "loss": 0.0815,
       "step": 951
     },
     {
       "epoch": 0.26,
       "learning_rate": 1.8233602374988404e-05,
-      "loss": 0.188,
+      "loss": 0.0805,
       "step": 952
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8231746915298266e-05,
-      "loss": 0.0365,
+      "loss": 0.0873,
       "step": 953
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8229891455608128e-05,
-      "loss": 0.194,
+      "loss": 0.252,
       "step": 954
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.822803599591799e-05,
-      "loss": 0.3407,
+      "loss": 0.247,
       "step": 955
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.822618053622785e-05,
-      "loss": 0.09,
+      "loss": 0.025,
       "step": 956
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8224325076537714e-05,
-      "loss": 0.1911,
+      "loss": 0.0842,
       "step": 957
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8222469616847575e-05,
-      "loss": 0.3917,
+      "loss": 0.1347,
       "step": 958
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8220614157157437e-05,
-      "loss": 0.1908,
+      "loss": 0.0277,
       "step": 959
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.82187586974673e-05,
-      "loss": 0.0416,
+      "loss": 0.191,
       "step": 960
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.821690323777716e-05,
-      "loss": 0.1943,
+      "loss": 0.0789,
       "step": 961
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8215047778087023e-05,
-      "loss": 0.2426,
+      "loss": 0.13,
       "step": 962
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8213192318396885e-05,
-      "loss": 0.0875,
+      "loss": 0.2965,
       "step": 963
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8211336858706747e-05,
-      "loss": 0.1412,
+      "loss": 0.1324,
       "step": 964
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.820948139901661e-05,
-      "loss": 0.2407,
+      "loss": 0.029,
       "step": 965
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.820762593932647e-05,
-      "loss": 0.0431,
+      "loss": 0.196,
       "step": 966
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8205770479636332e-05,
-      "loss": 0.1934,
+      "loss": 0.1401,
       "step": 967
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.820391501994619e-05,
-      "loss": 0.0962,
+      "loss": 0.0872,
       "step": 968
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8202059560256056e-05,
-      "loss": 0.0919,
+      "loss": 0.2414,
       "step": 969
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8200204100565918e-05,
-      "loss": 0.1399,
+      "loss": 0.0874,
       "step": 970
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.819834864087578e-05,
-      "loss": 0.0938,
+      "loss": 0.0879,
       "step": 971
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8196493181185638e-05,
-      "loss": 0.0895,
+      "loss": 0.1824,
       "step": 972
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8194637721495503e-05,
-      "loss": 0.0856,
+      "loss": 0.0821,
       "step": 973
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8192782261805365e-05,
-      "loss": 0.0881,
+      "loss": 0.0802,
       "step": 974
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8190926802115224e-05,
-      "loss": 0.0846,
+      "loss": 0.0802,
       "step": 975
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8189071342425086e-05,
-      "loss": 0.1936,
+      "loss": 0.1313,
       "step": 976
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.818721588273495e-05,
-      "loss": 0.2447,
+      "loss": 0.0789,
       "step": 977
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.818536042304481e-05,
-      "loss": 0.0864,
+      "loss": 0.028,
       "step": 978
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.818350496335467e-05,
-      "loss": 0.303,
+      "loss": 0.0802,
       "step": 979
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8181649503664533e-05,
-      "loss": 0.089,
+      "loss": 0.0856,
       "step": 980
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.81797940439744e-05,
-      "loss": 0.0837,
+      "loss": 0.0869,
       "step": 981
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8177938584284257e-05,
-      "loss": 0.0874,
+      "loss": 0.1436,
       "step": 982
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.817608312459412e-05,
-      "loss": 0.3066,
+      "loss": 0.0143,
       "step": 983
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.817422766490398e-05,
-      "loss": 0.3098,
+      "loss": 0.1402,
       "step": 984
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8172372205213843e-05,
-      "loss": 0.1375,
+      "loss": 0.0791,
       "step": 985
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8170516745523704e-05,
-      "loss": 0.2958,
+      "loss": 0.2636,
       "step": 986
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8168661285833566e-05,
-      "loss": 0.0318,
+      "loss": 0.1251,
       "step": 987
     },
     {
       "epoch": 0.27,
       "learning_rate": 1.8166805826143428e-05,
-      "loss": 0.0863,
+      "loss": 0.1362,
       "step": 988
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.816495036645329e-05,
-      "loss": 0.2467,
+      "loss": 0.2968,
       "step": 989
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8163094906763152e-05,
-      "loss": 0.0896,
+      "loss": 0.1884,
       "step": 990
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8161239447073014e-05,
-      "loss": 0.1404,
+      "loss": 0.1915,
       "step": 991
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8159383987382876e-05,
-      "loss": 0.1921,
+      "loss": 0.1416,
       "step": 992
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8157528527692737e-05,
-      "loss": 0.091,
+      "loss": 0.0928,
       "step": 993
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.81556730680026e-05,
-      "loss": 0.0881,
+      "loss": 0.1411,
       "step": 994
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.815381760831246e-05,
-      "loss": 0.0865,
+      "loss": 0.1421,
       "step": 995
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8151962148622323e-05,
-      "loss": 0.2485,
+      "loss": 0.1875,
       "step": 996
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8150106688932185e-05,
-      "loss": 0.0864,
+      "loss": 0.1871,
       "step": 997
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8148251229242047e-05,
-      "loss": 0.0374,
+      "loss": 0.1363,
       "step": 998
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.814639576955191e-05,
-      "loss": 0.2388,
+      "loss": 0.0888,
       "step": 999
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.814454030986177e-05,
-      "loss": 0.0863,
+      "loss": 0.2375,
       "step": 1000
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8142684850171632e-05,
-      "loss": 0.0853,
+      "loss": 0.093,
       "step": 1001
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8140829390481494e-05,
-      "loss": 0.085,
+      "loss": 0.1856,
       "step": 1002
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8138973930791356e-05,
-      "loss": 0.1392,
+      "loss": 0.1369,
       "step": 1003
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8137118471101215e-05,
-      "loss": 0.1388,
+      "loss": 0.0889,
       "step": 1004
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.813526301141108e-05,
-      "loss": 0.0328,
+      "loss": 0.1901,
       "step": 1005
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8133407551720942e-05,
-      "loss": 0.1372,
+      "loss": 0.0841,
       "step": 1006
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.81315520920308e-05,
-      "loss": 0.0312,
+      "loss": 0.138,
       "step": 1007
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8129696632340662e-05,
-      "loss": 0.2515,
+      "loss": 0.1322,
       "step": 1008
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8127841172650527e-05,
-      "loss": 0.2436,
+      "loss": 0.0276,
       "step": 1009
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.812598571296039e-05,
-      "loss": 0.0839,
+      "loss": 0.2401,
       "step": 1010
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8124130253270248e-05,
-      "loss": 0.0846,
+      "loss": 0.1968,
       "step": 1011
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.812227479358011e-05,
-      "loss": 0.1957,
+      "loss": 0.1423,
       "step": 1012
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8120419333889975e-05,
-      "loss": 0.3043,
+      "loss": 0.024,
       "step": 1013
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8118563874199833e-05,
-      "loss": 0.0857,
+      "loss": 0.0769,
       "step": 1014
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8116708414509695e-05,
-      "loss": 0.1886,
+      "loss": 0.0226,
       "step": 1015
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8114852954819557e-05,
-      "loss": 0.03,
+      "loss": 0.0871,
       "step": 1016
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8112997495129422e-05,
-      "loss": 0.0825,
+      "loss": 0.0725,
       "step": 1017
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.811114203543928e-05,
-      "loss": 0.0303,
+      "loss": 0.2588,
       "step": 1018
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8109286575749143e-05,
-      "loss": 0.3571,
+      "loss": 0.1431,
       "step": 1019
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8107431116059005e-05,
-      "loss": 0.0855,
+      "loss": 0.0756,
       "step": 1020
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8105575656368866e-05,
-      "loss": 0.192,
+      "loss": 0.0774,
       "step": 1021
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.810372019667873e-05,
-      "loss": 0.0834,
+      "loss": 0.1329,
       "step": 1022
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.810186473698859e-05,
-      "loss": 0.0845,
+      "loss": 0.0773,
       "step": 1023
     },
     {
       "epoch": 0.28,
       "learning_rate": 1.8100009277298452e-05,
-      "loss": 0.1951,
+      "loss": 0.0789,
       "step": 1024
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8098153817608314e-05,
-      "loss": 0.1997,
+      "loss": 0.2447,
       "step": 1025
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8096298357918176e-05,
-      "loss": 0.1412,
+      "loss": 0.0771,
       "step": 1026
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8094442898228038e-05,
-      "loss": 0.0329,
+      "loss": 0.0798,
       "step": 1027
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.80925874385379e-05,
-      "loss": 0.1397,
+      "loss": 0.0811,
       "step": 1028
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.809073197884776e-05,
-      "loss": 0.0316,
+      "loss": 0.0799,
       "step": 1029
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8088876519157623e-05,
-      "loss": 0.1924,
+      "loss": 0.2917,
       "step": 1030
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8087021059467485e-05,
-      "loss": 0.086,
+      "loss": 0.2344,
       "step": 1031
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8085165599777347e-05,
-      "loss": 0.1966,
+      "loss": 0.1851,
       "step": 1032
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.808331014008721e-05,
-      "loss": 0.2476,
+      "loss": 0.1379,
       "step": 1033
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.808145468039707e-05,
-      "loss": 0.1362,
+      "loss": 0.0459,
       "step": 1034
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8079599220706933e-05,
-      "loss": 0.0317,
+      "loss": 0.1415,
       "step": 1035
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.807774376101679e-05,
-      "loss": 0.1414,
+      "loss": 0.0455,
       "step": 1036
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8075888301326656e-05,
-      "loss": 0.2475,
+      "loss": 0.1806,
       "step": 1037
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8074032841636518e-05,
-      "loss": 0.1918,
+      "loss": 0.1811,
       "step": 1038
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.807217738194638e-05,
-      "loss": 0.1954,
+      "loss": 0.0839,
       "step": 1039
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.807032192225624e-05,
-      "loss": 0.1376,
+      "loss": 0.0843,
       "step": 1040
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8068466462566104e-05,
-      "loss": 0.0852,
+      "loss": 0.1312,
       "step": 1041
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8066611002875966e-05,
-      "loss": 0.1947,
+      "loss": 0.0254,
       "step": 1042
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8064755543185824e-05,
-      "loss": 0.1914,
+      "loss": 0.0226,
       "step": 1043
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8062900083495686e-05,
-      "loss": 0.0886,
+      "loss": 0.2051,
       "step": 1044
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.806104462380555e-05,
-      "loss": 0.1885,
+      "loss": 0.3291,
       "step": 1045
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8059189164115413e-05,
-      "loss": 0.1915,
+      "loss": 0.0195,
       "step": 1046
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8057333704425272e-05,
-      "loss": 0.1426,
+      "loss": 0.0201,
       "step": 1047
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8055478244735134e-05,
-      "loss": 0.0876,
+      "loss": 0.0724,
       "step": 1048
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8053622785045e-05,
-      "loss": 0.1903,
+      "loss": 0.1373,
       "step": 1049
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8051767325354857e-05,
-      "loss": 0.2436,
+      "loss": 0.0763,
       "step": 1050
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.804991186566472e-05,
-      "loss": 0.1894,
+      "loss": 0.0182,
       "step": 1051
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.804805640597458e-05,
-      "loss": 0.0889,
+      "loss": 0.0821,
       "step": 1052
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8046200946284443e-05,
-      "loss": 0.2373,
+      "loss": 0.0183,
       "step": 1053
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8044345486594305e-05,
-      "loss": 0.1332,
+      "loss": 0.0824,
       "step": 1054
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8042490026904167e-05,
-      "loss": 0.2386,
+      "loss": 0.1419,
       "step": 1055
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.804063456721403e-05,
-      "loss": 0.0977,
+      "loss": 0.195,
       "step": 1056
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.803877910752389e-05,
-      "loss": 0.1784,
+      "loss": 0.0785,
       "step": 1057
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8036923647833752e-05,
-      "loss": 0.192,
+      "loss": 0.1343,
       "step": 1058
     },
     {
       "epoch": 0.29,
       "learning_rate": 1.8035068188143614e-05,
-      "loss": 0.1904,
+      "loss": 0.0201,
       "step": 1059
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8033212728453476e-05,
-      "loss": 0.1881,
+      "loss": 0.0214,
       "step": 1060
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8031357268763338e-05,
-      "loss": 0.239,
+      "loss": 0.2466,
       "step": 1061
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.80295018090732e-05,
-      "loss": 0.1464,
+      "loss": 0.1396,
       "step": 1062
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.802764634938306e-05,
-      "loss": 0.1941,
+      "loss": 0.1286,
       "step": 1063
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8025790889692924e-05,
-      "loss": 0.4342,
+      "loss": 0.0825,
       "step": 1064
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8023935430002782e-05,
-      "loss": 0.1427,
+      "loss": 0.0794,
       "step": 1065
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8022079970312647e-05,
-      "loss": 0.1815,
+      "loss": 0.0783,
       "step": 1066
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.802022451062251e-05,
-      "loss": 0.1411,
+      "loss": 0.2462,
       "step": 1067
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.801836905093237e-05,
-      "loss": 0.1,
+      "loss": 0.0255,
       "step": 1068
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.801651359124223e-05,
-      "loss": 0.0994,
+      "loss": 0.0798,
       "step": 1069
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8014658131552095e-05,
-      "loss": 0.0972,
+      "loss": 0.1897,
       "step": 1070
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8012802671861957e-05,
-      "loss": 0.1933,
+      "loss": 0.1887,
       "step": 1071
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8010947212171815e-05,
-      "loss": 0.1417,
+      "loss": 0.0287,
       "step": 1072
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8009091752481677e-05,
-      "loss": 0.0918,
+      "loss": 0.076,
       "step": 1073
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8007236292791542e-05,
-      "loss": 0.1411,
+      "loss": 0.1313,
       "step": 1074
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8005380833101404e-05,
-      "loss": 0.0859,
+      "loss": 0.0768,
       "step": 1075
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8003525373411263e-05,
-      "loss": 0.1933,
+      "loss": 0.0775,
       "step": 1076
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.8001669913721124e-05,
-      "loss": 0.0843,
+      "loss": 0.0265,
       "step": 1077
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.799981445403099e-05,
-      "loss": 0.2553,
+      "loss": 0.0263,
       "step": 1078
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.7997958994340848e-05,
-      "loss": 0.1376,
+      "loss": 0.0752,
       "step": 1079
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.799610353465071e-05,
-      "loss": 0.0849,
+      "loss": 0.0216,
       "step": 1080
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.7994248074960572e-05,
-      "loss": 0.0837,
+      "loss": 0.0757,
       "step": 1081
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.7992392615270434e-05,
-      "loss": 0.1959,
+      "loss": 0.1431,
       "step": 1082
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.7990537155580296e-05,
-      "loss": 0.0856,
+      "loss": 0.1442,
       "step": 1083
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.7988681695890158e-05,
-      "loss": 0.2466,
+      "loss": 0.0149,
       "step": 1084
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.798682623620002e-05,
-      "loss": 0.1382,
+      "loss": 0.1995,
       "step": 1085
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.798497077650988e-05,
-      "loss": 0.1938,
+      "loss": 0.2007,
       "step": 1086
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.7983115316819743e-05,
-      "loss": 0.1949,
+      "loss": 0.075,
       "step": 1087
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.7981259857129605e-05,
-      "loss": 0.1923,
+      "loss": 0.3128,
       "step": 1088
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.7979404397439467e-05,
-      "loss": 0.1413,
+      "loss": 0.0208,
       "step": 1089
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.797754893774933e-05,
-      "loss": 0.1346,
+      "loss": 0.025,
       "step": 1090
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.797569347805919e-05,
-      "loss": 0.1353,
+      "loss": 0.1931,
       "step": 1091
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.7973838018369052e-05,
-      "loss": 0.1912,
+      "loss": 0.0767,
       "step": 1092
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.7971982558678914e-05,
-      "loss": 0.0857,
+      "loss": 0.1354,
       "step": 1093
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.7970127098988776e-05,
-      "loss": 0.0318,
+      "loss": 0.1311,
       "step": 1094
     },
     {
       "epoch": 0.3,
       "learning_rate": 1.7968271639298638e-05,
-      "loss": 0.1332,
+      "loss": 0.2417,
       "step": 1095
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.79664161796085e-05,
-      "loss": 0.1391,
+      "loss": 0.0805,
       "step": 1096
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7964560719918362e-05,
-      "loss": 0.1962,
+      "loss": 0.0318,
       "step": 1097
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7962705260228224e-05,
-      "loss": 0.1872,
+      "loss": 0.2928,
       "step": 1098
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7960849800538086e-05,
-      "loss": 0.1385,
+      "loss": 0.1371,
       "step": 1099
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7958994340847947e-05,
-      "loss": 0.1407,
+      "loss": 0.0854,
       "step": 1100
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7957138881157806e-05,
-      "loss": 0.0878,
+      "loss": 0.1361,
       "step": 1101
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.795528342146767e-05,
-      "loss": 0.1909,
+      "loss": 0.1384,
       "step": 1102
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7953427961777533e-05,
-      "loss": 0.1344,
+      "loss": 0.035,
       "step": 1103
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.795157250208739e-05,
-      "loss": 0.0331,
+      "loss": 0.0366,
       "step": 1104
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7949717042397253e-05,
-      "loss": 0.0885,
+      "loss": 0.1888,
       "step": 1105
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.794786158270712e-05,
-      "loss": 0.1917,
+      "loss": 0.0856,
       "step": 1106
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.794600612301698e-05,
-      "loss": 0.0842,
+      "loss": 0.0303,
       "step": 1107
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.794415066332684e-05,
-      "loss": 0.1915,
+      "loss": 0.0833,
       "step": 1108
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.79422952036367e-05,
-      "loss": 0.1923,
+      "loss": 0.0823,
       "step": 1109
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7940439743946566e-05,
-      "loss": 0.0308,
+      "loss": 0.1354,
       "step": 1110
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7938584284256425e-05,
-      "loss": 0.2495,
+      "loss": 0.0754,
       "step": 1111
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7936728824566287e-05,
-      "loss": 0.0333,
+      "loss": 0.0781,
       "step": 1112
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.793487336487615e-05,
-      "loss": 0.0853,
+      "loss": 0.0205,
       "step": 1113
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7933017905186014e-05,
-      "loss": 0.1961,
+      "loss": 0.2015,
       "step": 1114
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7931162445495872e-05,
-      "loss": 0.1387,
+      "loss": 0.0763,
       "step": 1115
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7929306985805734e-05,
-      "loss": 0.1383,
+      "loss": 0.1926,
       "step": 1116
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7927451526115596e-05,
-      "loss": 0.3598,
+      "loss": 0.0176,
       "step": 1117
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7925596066425458e-05,
-      "loss": 0.2522,
+      "loss": 0.1266,
       "step": 1118
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.792374060673532e-05,
-      "loss": 0.0897,
+      "loss": 0.0768,
       "step": 1119
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.792188514704518e-05,
-      "loss": 0.0823,
+      "loss": 0.1383,
       "step": 1120
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7920029687355043e-05,
-      "loss": 0.1441,
+      "loss": 0.083,
       "step": 1121
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7918174227664905e-05,
-      "loss": 0.1366,
+      "loss": 0.1289,
       "step": 1122
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7916318767974767e-05,
-      "loss": 0.192,
+      "loss": 0.0192,
       "step": 1123
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.791446330828463e-05,
-      "loss": 0.3465,
+      "loss": 0.2501,
       "step": 1124
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.791260784859449e-05,
-      "loss": 0.0869,
+      "loss": 0.0823,
       "step": 1125
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7910752388904353e-05,
-      "loss": 0.0882,
+      "loss": 0.305,
       "step": 1126
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7908896929214215e-05,
-      "loss": 0.1413,
+      "loss": 0.0766,
       "step": 1127
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7907041469524076e-05,
-      "loss": 0.2525,
+      "loss": 0.083,
       "step": 1128
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.790518600983394e-05,
-      "loss": 0.1894,
+      "loss": 0.0788,
       "step": 1129
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.79033305501438e-05,
-      "loss": 0.1894,
+      "loss": 0.0797,
       "step": 1130
     },
     {
       "epoch": 0.31,
       "learning_rate": 1.7901475090453662e-05,
-      "loss": 0.2415,
+      "loss": 0.0301,
       "step": 1131
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7899619630763524e-05,
-      "loss": 0.246,
+      "loss": 0.1369,
       "step": 1132
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7897764171073382e-05,
-      "loss": 0.0945,
+      "loss": 0.0252,
       "step": 1133
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7895908711383248e-05,
-      "loss": 0.1424,
+      "loss": 0.0784,
       "step": 1134
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.789405325169311e-05,
-      "loss": 0.1378,
+      "loss": 0.0795,
       "step": 1135
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.789219779200297e-05,
-      "loss": 0.2406,
+      "loss": 0.1384,
       "step": 1136
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.789034233231283e-05,
-      "loss": 0.1426,
+      "loss": 0.0736,
       "step": 1137
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7888486872622695e-05,
-      "loss": 0.1918,
+      "loss": 0.0208,
       "step": 1138
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7886631412932557e-05,
-      "loss": 0.093,
+      "loss": 0.0795,
       "step": 1139
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7884775953242416e-05,
-      "loss": 0.1408,
+      "loss": 0.1336,
       "step": 1140
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7882920493552277e-05,
-      "loss": 0.0926,
+      "loss": 0.1389,
       "step": 1141
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7881065033862143e-05,
-      "loss": 0.1384,
+      "loss": 0.1344,
       "step": 1142
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7879209574172005e-05,
-      "loss": 0.3924,
+      "loss": 0.0763,
       "step": 1143
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7877354114481863e-05,
-      "loss": 0.0951,
+      "loss": 0.0162,
       "step": 1144
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7875498654791725e-05,
-      "loss": 0.0957,
+      "loss": 0.3099,
       "step": 1145
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.787364319510159e-05,
-      "loss": 0.0922,
+      "loss": 0.2566,
       "step": 1146
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.787178773541145e-05,
-      "loss": 0.1394,
+      "loss": 0.1436,
       "step": 1147
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.786993227572131e-05,
-      "loss": 0.0377,
+      "loss": 0.1364,
       "step": 1148
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7868076816031172e-05,
-      "loss": 0.0884,
+      "loss": 0.0796,
       "step": 1149
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7866221356341034e-05,
-      "loss": 0.0369,
+      "loss": 0.0341,
       "step": 1150
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7864365896650896e-05,
-      "loss": 0.1904,
+      "loss": 0.0307,
       "step": 1151
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7862510436960758e-05,
-      "loss": 0.1424,
+      "loss": 0.1322,
       "step": 1152
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.786065497727062e-05,
-      "loss": 0.1984,
+      "loss": 0.0346,
       "step": 1153
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.785879951758048e-05,
-      "loss": 0.0844,
+      "loss": 0.0794,
       "step": 1154
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7856944057890344e-05,
-      "loss": 0.1362,
+      "loss": 0.1318,
       "step": 1155
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7855088598200205e-05,
-      "loss": 0.0859,
+      "loss": 0.0823,
       "step": 1156
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7853233138510067e-05,
-      "loss": 0.1866,
+      "loss": 0.1343,
       "step": 1157
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.785137767881993e-05,
-      "loss": 0.0799,
+      "loss": 0.0245,
       "step": 1158
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.784952221912979e-05,
-      "loss": 0.1368,
+      "loss": 0.1255,
       "step": 1159
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7847666759439653e-05,
-      "loss": 0.3598,
+      "loss": 0.024,
       "step": 1160
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7845811299749515e-05,
-      "loss": 0.1962,
+      "loss": 0.0748,
       "step": 1161
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7843955840059377e-05,
-      "loss": 0.2437,
+      "loss": 0.185,
       "step": 1162
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.784210038036924e-05,
-      "loss": 0.0281,
+      "loss": 0.1965,
       "step": 1163
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.78402449206791e-05,
-      "loss": 0.1956,
+      "loss": 0.1365,
       "step": 1164
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7838389460988962e-05,
-      "loss": 0.0848,
+      "loss": 0.1884,
       "step": 1165
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7836534001298824e-05,
-      "loss": 0.1884,
+      "loss": 0.136,
       "step": 1166
     },
     {
       "epoch": 0.32,
       "learning_rate": 1.7834678541608686e-05,
-      "loss": 0.2498,
+      "loss": 0.1887,
       "step": 1167
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7832823081918548e-05,
-      "loss": 0.0319,
+      "loss": 0.0781,
       "step": 1168
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7830967622228406e-05,
-      "loss": 0.0309,
+      "loss": 0.0236,
       "step": 1169
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.782911216253827e-05,
-      "loss": 0.0309,
+      "loss": 0.1377,
       "step": 1170
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7827256702848133e-05,
-      "loss": 0.1973,
+      "loss": 0.0793,
       "step": 1171
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7825401243157995e-05,
-      "loss": 0.1915,
+      "loss": 0.1307,
       "step": 1172
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7823545783467854e-05,
-      "loss": 0.0864,
+      "loss": 0.1847,
       "step": 1173
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.782169032377772e-05,
-      "loss": 0.1354,
+      "loss": 0.0771,
       "step": 1174
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.781983486408758e-05,
-      "loss": 0.1382,
+      "loss": 0.0284,
       "step": 1175
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.781797940439744e-05,
-      "loss": 0.0859,
+      "loss": 0.0787,
       "step": 1176
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.78161239447073e-05,
-      "loss": 0.25,
+      "loss": 0.0788,
       "step": 1177
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7814268485017167e-05,
-      "loss": 0.1371,
+      "loss": 0.084,
       "step": 1178
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7812413025327025e-05,
-      "loss": 0.4597,
+      "loss": 0.1904,
       "step": 1179
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7810557565636887e-05,
-      "loss": 0.1415,
+      "loss": 0.1326,
       "step": 1180
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.780870210594675e-05,
-      "loss": 0.2475,
+      "loss": 0.0826,
       "step": 1181
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7806846646256614e-05,
-      "loss": 0.196,
+      "loss": 0.0293,
       "step": 1182
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7804991186566473e-05,
-      "loss": 0.1379,
+      "loss": 0.0246,
       "step": 1183
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7803135726876334e-05,
-      "loss": 0.1941,
+      "loss": 0.1342,
       "step": 1184
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7801280267186196e-05,
-      "loss": 0.0893,
+      "loss": 0.0236,
       "step": 1185
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7799424807496058e-05,
-      "loss": 0.2392,
+      "loss": 0.2385,
       "step": 1186
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.779756934780592e-05,
-      "loss": 0.0889,
+      "loss": 0.1929,
       "step": 1187
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7795713888115782e-05,
-      "loss": 0.1412,
+      "loss": 0.0199,
       "step": 1188
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7793858428425644e-05,
-      "loss": 0.1939,
+      "loss": 0.0242,
       "step": 1189
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7792002968735506e-05,
-      "loss": 0.0381,
+      "loss": 0.0763,
       "step": 1190
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7790147509045368e-05,
-      "loss": 0.1414,
+      "loss": 0.0814,
       "step": 1191
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.778829204935523e-05,
-      "loss": 0.1374,
+      "loss": 0.0193,
       "step": 1192
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.778643658966509e-05,
-      "loss": 0.2437,
+      "loss": 0.1356,
       "step": 1193
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7784581129974953e-05,
-      "loss": 0.2377,
+      "loss": 0.1417,
       "step": 1194
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7782725670284815e-05,
-      "loss": 0.1368,
+      "loss": 0.0807,
       "step": 1195
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7780870210594677e-05,
-      "loss": 0.1906,
+      "loss": 0.0177,
       "step": 1196
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.777901475090454e-05,
-      "loss": 0.0892,
+      "loss": 0.1327,
       "step": 1197
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.77771592912144e-05,
-      "loss": 0.0888,
+      "loss": 0.1407,
       "step": 1198
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7775303831524262e-05,
-      "loss": 0.0905,
+      "loss": 0.0782,
       "step": 1199
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7773448371834124e-05,
-      "loss": 0.1359,
+      "loss": 0.203,
       "step": 1200
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7771592912143983e-05,
-      "loss": 0.2408,
+      "loss": 0.1365,
       "step": 1201
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.7769737452453848e-05,
-      "loss": 0.2908,
+      "loss": 0.198,
       "step": 1202
     },
     {
       "epoch": 0.33,
       "learning_rate": 1.776788199276371e-05,
-      "loss": 0.1901,
+      "loss": 0.0221,
       "step": 1203
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7766026533073572e-05,
-      "loss": 0.1933,
+      "loss": 0.1846,
       "step": 1204
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.776417107338343e-05,
-      "loss": 0.2366,
+      "loss": 0.0194,
       "step": 1205
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7762315613693296e-05,
-      "loss": 0.1909,
+      "loss": 0.1855,
       "step": 1206
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7760460154003157e-05,
-      "loss": 0.0888,
+      "loss": 0.0846,
       "step": 1207
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7758604694313016e-05,
-      "loss": 0.0925,
+      "loss": 0.1343,
       "step": 1208
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7756749234622878e-05,
-      "loss": 0.2416,
+      "loss": 0.1363,
       "step": 1209
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7754893774932743e-05,
-      "loss": 0.244,
+      "loss": 0.0834,
       "step": 1210
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7753038315242605e-05,
-      "loss": 0.2316,
+      "loss": 0.1339,
       "step": 1211
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7751182855552463e-05,
-      "loss": 0.0942,
+      "loss": 0.1794,
       "step": 1212
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7749327395862325e-05,
-      "loss": 0.0439,
+      "loss": 0.0831,
       "step": 1213
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.774747193617219e-05,
-      "loss": 0.1403,
+      "loss": 0.1807,
       "step": 1214
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.774561647648205e-05,
-      "loss": 0.1982,
+      "loss": 0.1384,
       "step": 1215
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.774376101679191e-05,
-      "loss": 0.0941,
+      "loss": 0.1785,
       "step": 1216
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7741905557101773e-05,
-      "loss": 0.0422,
+      "loss": 0.0878,
       "step": 1217
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7740050097411635e-05,
-      "loss": 0.09,
+      "loss": 0.1309,
       "step": 1218
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7738194637721497e-05,
-      "loss": 0.1421,
+      "loss": 0.1308,
       "step": 1219
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.773633917803136e-05,
-      "loss": 0.1911,
+      "loss": 0.0397,
       "step": 1220
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.773448371834122e-05,
-      "loss": 0.1426,
+      "loss": 0.1245,
       "step": 1221
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7732628258651082e-05,
-      "loss": 0.1366,
+      "loss": 0.0855,
       "step": 1222
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7730772798960944e-05,
-      "loss": 0.1417,
+      "loss": 0.1902,
       "step": 1223
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7728917339270806e-05,
-      "loss": 0.1887,
+      "loss": 0.1353,
       "step": 1224
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7727061879580668e-05,
-      "loss": 0.1932,
+      "loss": 0.0213,
       "step": 1225
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.772520641989053e-05,
-      "loss": 0.1937,
+      "loss": 0.1908,
       "step": 1226
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.772335096020039e-05,
-      "loss": 0.2463,
+      "loss": 0.0711,
       "step": 1227
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7721495500510253e-05,
-      "loss": 0.1356,
+      "loss": 0.1959,
       "step": 1228
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7719640040820115e-05,
-      "loss": 0.1827,
+      "loss": 0.0788,
       "step": 1229
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7717784581129974e-05,
-      "loss": 0.0874,
+      "loss": 0.0169,
       "step": 1230
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.771592912143984e-05,
-      "loss": 0.091,
+      "loss": 0.0719,
       "step": 1231
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.77140736617497e-05,
-      "loss": 0.3011,
+      "loss": 0.257,
       "step": 1232
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7712218202059563e-05,
-      "loss": 0.0845,
+      "loss": 0.1941,
       "step": 1233
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.771036274236942e-05,
-      "loss": 0.139,
+      "loss": 0.1381,
       "step": 1234
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7708507282679286e-05,
-      "loss": 0.2438,
+      "loss": 0.0799,
       "step": 1235
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7706651822989148e-05,
-      "loss": 0.088,
+      "loss": 0.1282,
       "step": 1236
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7704796363299007e-05,
-      "loss": 0.1418,
+      "loss": 0.1374,
       "step": 1237
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.770294090360887e-05,
-      "loss": 0.1445,
+      "loss": 0.1448,
       "step": 1238
     },
     {
       "epoch": 0.34,
       "learning_rate": 1.7701085443918734e-05,
-      "loss": 0.2966,
+      "loss": 0.1273,
       "step": 1239
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7699229984228596e-05,
-      "loss": 0.0358,
+      "loss": 0.136,
       "step": 1240
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7697374524538454e-05,
-      "loss": 0.1924,
+      "loss": 0.1249,
       "step": 1241
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7695519064848316e-05,
-      "loss": 0.138,
+      "loss": 0.0971,
       "step": 1242
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.769366360515818e-05,
-      "loss": 0.1975,
+      "loss": 0.0813,
       "step": 1243
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.769180814546804e-05,
-      "loss": 0.036,
+      "loss": 0.0855,
       "step": 1244
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7689952685777902e-05,
-      "loss": 0.291,
+      "loss": 0.0787,
       "step": 1245
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7688097226087764e-05,
-      "loss": 0.2417,
+      "loss": 0.0317,
       "step": 1246
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7686241766397625e-05,
-      "loss": 0.1437,
+      "loss": 0.0325,
       "step": 1247
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7684386306707487e-05,
-      "loss": 0.1926,
+      "loss": 0.1297,
       "step": 1248
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.768253084701735e-05,
-      "loss": 0.0913,
+      "loss": 0.135,
       "step": 1249
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.768067538732721e-05,
-      "loss": 0.1424,
+      "loss": 0.1939,
       "step": 1250
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7678819927637073e-05,
-      "loss": 0.1395,
+      "loss": 0.1853,
       "step": 1251
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7676964467946935e-05,
-      "loss": 0.0384,
+      "loss": 0.1401,
       "step": 1252
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7675109008256797e-05,
-      "loss": 0.138,
+      "loss": 0.0787,
       "step": 1253
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.767325354856666e-05,
-      "loss": 0.0894,
+      "loss": 0.1363,
       "step": 1254
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.767139808887652e-05,
-      "loss": 0.2882,
+      "loss": 0.1344,
       "step": 1255
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7669542629186382e-05,
-      "loss": 0.1913,
+      "loss": 0.1911,
       "step": 1256
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7667687169496244e-05,
-      "loss": 0.1377,
+      "loss": 0.0219,
       "step": 1257
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7665831709806106e-05,
-      "loss": 0.0852,
+      "loss": 0.0783,
       "step": 1258
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7663976250115968e-05,
-      "loss": 0.0852,
+      "loss": 0.0757,
       "step": 1259
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.766212079042583e-05,
-      "loss": 0.1409,
+      "loss": 0.2566,
       "step": 1260
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.766026533073569e-05,
-      "loss": 0.0861,
+      "loss": 0.1289,
       "step": 1261
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7658409871045554e-05,
-      "loss": 0.139,
+      "loss": 0.0219,
       "step": 1262
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7656554411355415e-05,
-      "loss": 0.2448,
+      "loss": 0.0781,
       "step": 1263
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7654698951665277e-05,
-      "loss": 0.145,
+      "loss": 0.1361,
       "step": 1264
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.765284349197514e-05,
-      "loss": 0.0835,
+      "loss": 0.082,
       "step": 1265
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7650988032284998e-05,
-      "loss": 0.1918,
+      "loss": 0.0787,
       "step": 1266
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7649132572594863e-05,
-      "loss": 0.2483,
+      "loss": 0.021,
       "step": 1267
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7647277112904725e-05,
-      "loss": 0.2427,
+      "loss": 0.0204,
       "step": 1268
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7645421653214587e-05,
-      "loss": 0.1874,
+      "loss": 0.0709,
       "step": 1269
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7643566193524445e-05,
-      "loss": 0.2972,
+      "loss": 0.014,
       "step": 1270
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.764171073383431e-05,
-      "loss": 0.0878,
+      "loss": 0.1358,
       "step": 1271
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7639855274144172e-05,
-      "loss": 0.1952,
+      "loss": 0.007,
       "step": 1272
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.763799981445403e-05,
-      "loss": 0.1928,
+      "loss": 0.0062,
       "step": 1273
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7636144354763893e-05,
-      "loss": 0.1378,
+      "loss": 0.0044,
       "step": 1274
     },
     {
       "epoch": 0.35,
       "learning_rate": 1.7634288895073758e-05,
-      "loss": 0.1387,
+      "loss": 0.0949,
       "step": 1275
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7632433435383616e-05,
-      "loss": 0.0404,
+      "loss": 0.3541,
       "step": 1276
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7630577975693478e-05,
-      "loss": 0.138,
+      "loss": 0.1746,
       "step": 1277
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.762872251600334e-05,
-      "loss": 0.0389,
+      "loss": 0.0923,
       "step": 1278
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7626867056313205e-05,
-      "loss": 0.2483,
+      "loss": 0.0055,
       "step": 1279
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7625011596623064e-05,
-      "loss": 0.1414,
+      "loss": 0.0858,
       "step": 1280
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7623156136932926e-05,
-      "loss": 0.0899,
+      "loss": 0.2299,
       "step": 1281
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7621300677242788e-05,
-      "loss": 0.1863,
+      "loss": 0.1442,
       "step": 1282
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.761944521755265e-05,
-      "loss": 0.0385,
+      "loss": 0.1403,
       "step": 1283
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.761758975786251e-05,
-      "loss": 0.0893,
+      "loss": 0.1013,
       "step": 1284
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7615734298172373e-05,
-      "loss": 0.3027,
+      "loss": 0.0908,
       "step": 1285
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7613878838482235e-05,
-      "loss": 0.1412,
+      "loss": 0.1436,
       "step": 1286
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7612023378792097e-05,
-      "loss": 0.1911,
+      "loss": 0.0812,
       "step": 1287
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.761016791910196e-05,
-      "loss": 0.2449,
+      "loss": 0.0168,
       "step": 1288
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.760831245941182e-05,
-      "loss": 0.1437,
+      "loss": 0.0812,
       "step": 1289
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7606456999721683e-05,
-      "loss": 0.0849,
+      "loss": 0.1375,
       "step": 1290
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7604601540031544e-05,
-      "loss": 0.2348,
+      "loss": 0.0138,
       "step": 1291
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7602746080341406e-05,
-      "loss": 0.1393,
+      "loss": 0.3942,
       "step": 1292
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7600890620651268e-05,
-      "loss": 0.0888,
+      "loss": 0.0807,
       "step": 1293
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.759903516096113e-05,
-      "loss": 0.1413,
+      "loss": 0.0809,
       "step": 1294
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7597179701270992e-05,
-      "loss": 0.0867,
+      "loss": 0.0834,
       "step": 1295
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7595324241580854e-05,
-      "loss": 0.142,
+      "loss": 0.219,
       "step": 1296
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7593468781890716e-05,
-      "loss": 0.1422,
+      "loss": 0.0803,
       "step": 1297
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7591613322200578e-05,
-      "loss": 0.1891,
+      "loss": 0.1471,
       "step": 1298
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.758975786251044e-05,
-      "loss": 0.0335,
+      "loss": 0.0868,
       "step": 1299
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.75879024028203e-05,
-      "loss": 0.145,
+      "loss": 0.0821,
       "step": 1300
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7586046943130163e-05,
-      "loss": 0.0337,
+      "loss": 0.0259,
       "step": 1301
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.758419148344002e-05,
-      "loss": 0.1346,
+      "loss": 0.0804,
       "step": 1302
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7582336023749887e-05,
-      "loss": 0.2978,
+      "loss": 0.023,
       "step": 1303
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.758048056405975e-05,
-      "loss": 0.3469,
+      "loss": 0.0794,
       "step": 1304
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7578625104369607e-05,
-      "loss": 0.2396,
+      "loss": 0.0169,
       "step": 1305
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.757676964467947e-05,
-      "loss": 0.2974,
+      "loss": 0.0849,
       "step": 1306
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7574914184989334e-05,
-      "loss": 0.1425,
+      "loss": 0.2749,
       "step": 1307
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7573058725299196e-05,
-      "loss": 0.1403,
+      "loss": 0.1547,
       "step": 1308
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7571203265609055e-05,
-      "loss": 0.1421,
+      "loss": 0.1431,
       "step": 1309
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7569347805918917e-05,
-      "loss": 0.1402,
+      "loss": 0.1431,
       "step": 1310
     },
     {
       "epoch": 0.36,
       "learning_rate": 1.7567492346228782e-05,
-      "loss": 0.1364,
+      "loss": 0.0845,
       "step": 1311
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.756563688653864e-05,
-      "loss": 0.0379,
+      "loss": 0.0208,
       "step": 1312
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7563781426848502e-05,
-      "loss": 0.0862,
+      "loss": 0.2112,
       "step": 1313
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7561925967158364e-05,
-      "loss": 0.0896,
+      "loss": 0.0234,
       "step": 1314
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.756007050746823e-05,
-      "loss": 0.1974,
+      "loss": 0.079,
       "step": 1315
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7558215047778088e-05,
-      "loss": 0.1891,
+      "loss": 0.0228,
       "step": 1316
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.755635958808795e-05,
-      "loss": 0.0869,
+      "loss": 0.0211,
       "step": 1317
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.755450412839781e-05,
-      "loss": 0.0883,
+      "loss": 0.0816,
       "step": 1318
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7552648668707673e-05,
-      "loss": 0.2375,
+      "loss": 0.0185,
       "step": 1319
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7550793209017535e-05,
-      "loss": 0.1388,
+      "loss": 0.0814,
       "step": 1320
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7548937749327397e-05,
-      "loss": 0.1391,
+      "loss": 0.2773,
       "step": 1321
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.754708228963726e-05,
-      "loss": 0.0366,
+      "loss": 0.0775,
       "step": 1322
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.754522682994712e-05,
-      "loss": 0.1925,
+      "loss": 0.0143,
       "step": 1323
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7543371370256983e-05,
-      "loss": 0.291,
+      "loss": 0.2197,
       "step": 1324
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7541515910566845e-05,
-      "loss": 0.346,
+      "loss": 0.3345,
       "step": 1325
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7539660450876706e-05,
-      "loss": 0.1427,
+      "loss": 0.1504,
       "step": 1326
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.753780499118657e-05,
-      "loss": 0.0377,
+      "loss": 0.018,
       "step": 1327
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.753594953149643e-05,
-      "loss": 0.0858,
+      "loss": 0.0777,
       "step": 1328
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7534094071806292e-05,
-      "loss": 0.0903,
+      "loss": 0.1307,
       "step": 1329
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7532238612116154e-05,
-      "loss": 0.0861,
+      "loss": 0.1412,
       "step": 1330
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7530383152426016e-05,
-      "loss": 0.09,
+      "loss": 0.1896,
       "step": 1331
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7528527692735878e-05,
-      "loss": 0.0847,
+      "loss": 0.0797,
       "step": 1332
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.752667223304574e-05,
-      "loss": 0.1362,
+      "loss": 0.082,
       "step": 1333
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7524816773355598e-05,
-      "loss": 0.2971,
+      "loss": 0.0837,
       "step": 1334
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7522961313665463e-05,
-      "loss": 0.2508,
+      "loss": 0.077,
       "step": 1335
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7521105853975325e-05,
-      "loss": 0.0874,
+      "loss": 0.1364,
       "step": 1336
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7519250394285187e-05,
-      "loss": 0.0833,
+      "loss": 0.1915,
       "step": 1337
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7517394934595046e-05,
-      "loss": 0.0862,
+      "loss": 0.1907,
       "step": 1338
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.751553947490491e-05,
-      "loss": 0.1907,
+      "loss": 0.2903,
       "step": 1339
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7513684015214773e-05,
-      "loss": 0.1913,
+      "loss": 0.0854,
       "step": 1340
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.751182855552463e-05,
-      "loss": 0.2512,
+      "loss": 0.1792,
       "step": 1341
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7509973095834493e-05,
-      "loss": 0.1433,
+      "loss": 0.0832,
       "step": 1342
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7508117636144358e-05,
-      "loss": 0.2425,
+      "loss": 0.1809,
       "step": 1343
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7506262176454217e-05,
-      "loss": 0.2391,
+      "loss": 0.1443,
       "step": 1344
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.750440671676408e-05,
-      "loss": 0.2453,
+      "loss": 0.2293,
       "step": 1345
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.750255125707394e-05,
-      "loss": 0.1338,
+      "loss": 0.0405,
       "step": 1346
     },
     {
       "epoch": 0.37,
       "learning_rate": 1.7500695797383806e-05,
-      "loss": 0.1915,
+      "loss": 0.0828,
       "step": 1347
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7498840337693664e-05,
-      "loss": 0.1356,
+      "loss": 0.1307,
       "step": 1348
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7496984878003526e-05,
-      "loss": 0.1905,
+      "loss": 0.3146,
       "step": 1349
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7495129418313388e-05,
-      "loss": 0.1406,
+      "loss": 0.0837,
       "step": 1350
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.749327395862325e-05,
-      "loss": 0.1444,
+      "loss": 0.1343,
       "step": 1351
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7491418498933112e-05,
-      "loss": 0.2922,
+      "loss": 0.0353,
       "step": 1352
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7489563039242974e-05,
-      "loss": 0.0423,
+      "loss": 0.1322,
       "step": 1353
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7487707579552835e-05,
-      "loss": 0.3366,
+      "loss": 0.0793,
       "step": 1354
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7485852119862697e-05,
-      "loss": 0.3291,
+      "loss": 0.1999,
       "step": 1355
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.748399666017256e-05,
-      "loss": 0.0931,
+      "loss": 0.0194,
       "step": 1356
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.748214120048242e-05,
-      "loss": 0.142,
+      "loss": 0.1446,
       "step": 1357
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7480285740792283e-05,
-      "loss": 0.0958,
+      "loss": 0.207,
       "step": 1358
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7478430281102145e-05,
-      "loss": 0.1909,
+      "loss": 0.1922,
       "step": 1359
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7476574821412007e-05,
-      "loss": 0.3269,
+      "loss": 0.2349,
       "step": 1360
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.747471936172187e-05,
-      "loss": 0.1394,
+      "loss": 0.0853,
       "step": 1361
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.747286390203173e-05,
-      "loss": 0.1397,
+      "loss": 0.0828,
       "step": 1362
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7471008442341592e-05,
-      "loss": 0.1888,
+      "loss": 0.2363,
       "step": 1363
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7469152982651454e-05,
-      "loss": 0.2318,
+      "loss": 0.0918,
       "step": 1364
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7467297522961316e-05,
-      "loss": 0.2003,
+      "loss": 0.2261,
       "step": 1365
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7465442063271178e-05,
-      "loss": 0.1403,
+      "loss": 0.1706,
       "step": 1366
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7463586603581036e-05,
-      "loss": 0.0986,
+      "loss": 0.173,
       "step": 1367
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.74617311438909e-05,
-      "loss": 0.2398,
+      "loss": 0.0538,
       "step": 1368
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7459875684200764e-05,
-      "loss": 0.1971,
+      "loss": 0.052,
       "step": 1369
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7458020224510622e-05,
-      "loss": 0.1455,
+      "loss": 0.1409,
       "step": 1370
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7456164764820484e-05,
-      "loss": 0.1903,
+      "loss": 0.0422,
       "step": 1371
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.745430930513035e-05,
-      "loss": 0.1931,
+      "loss": 0.0874,
       "step": 1372
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7452453845440208e-05,
-      "loss": 0.1423,
+      "loss": 0.1252,
       "step": 1373
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.745059838575007e-05,
-      "loss": 0.0957,
+      "loss": 0.2918,
       "step": 1374
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.744874292605993e-05,
-      "loss": 0.0949,
+      "loss": 0.0748,
       "step": 1375
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7446887466369797e-05,
-      "loss": 0.0921,
+      "loss": 0.1348,
       "step": 1376
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7445032006679655e-05,
-      "loss": 0.1944,
+      "loss": 0.0199,
       "step": 1377
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7443176546989517e-05,
-      "loss": 0.1357,
+      "loss": 0.073,
       "step": 1378
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.744132108729938e-05,
-      "loss": 0.2351,
+      "loss": 0.1887,
       "step": 1379
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.743946562760924e-05,
-      "loss": 0.193,
+      "loss": 0.184,
       "step": 1380
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7437610167919103e-05,
-      "loss": 0.1371,
+      "loss": 0.0173,
       "step": 1381
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7435754708228964e-05,
-      "loss": 0.2388,
+      "loss": 0.0164,
       "step": 1382
     },
     {
       "epoch": 0.38,
       "learning_rate": 1.7433899248538826e-05,
-      "loss": 0.1395,
+      "loss": 0.0714,
       "step": 1383
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7432043788848688e-05,
-      "loss": 0.0901,
+      "loss": 0.1329,
       "step": 1384
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.743018832915855e-05,
-      "loss": 0.1435,
+      "loss": 0.1284,
       "step": 1385
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7428332869468412e-05,
-      "loss": 0.2861,
+      "loss": 0.0204,
       "step": 1386
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7426477409778274e-05,
-      "loss": 0.0852,
+      "loss": 0.0203,
       "step": 1387
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7424621950088136e-05,
-      "loss": 0.1935,
+      "loss": 0.087,
       "step": 1388
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7422766490397998e-05,
-      "loss": 0.0949,
+      "loss": 0.0234,
       "step": 1389
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.742091103070786e-05,
-      "loss": 0.1377,
+      "loss": 0.1254,
       "step": 1390
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.741905557101772e-05,
-      "loss": 0.0909,
+      "loss": 0.0268,
       "step": 1391
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7417200111327583e-05,
-      "loss": 0.0374,
+      "loss": 0.0697,
       "step": 1392
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7415344651637445e-05,
-      "loss": 0.1906,
+      "loss": 0.0173,
       "step": 1393
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7413489191947307e-05,
-      "loss": 0.0367,
+      "loss": 0.0101,
       "step": 1394
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.741163373225717e-05,
-      "loss": 0.2408,
+      "loss": 0.0833,
       "step": 1395
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.740977827256703e-05,
-      "loss": 0.1398,
+      "loss": 0.2453,
       "step": 1396
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7407922812876893e-05,
-      "loss": 0.2465,
+      "loss": 0.0092,
       "step": 1397
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7406067353186754e-05,
-      "loss": 0.1921,
+      "loss": 0.0095,
       "step": 1398
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7404211893496613e-05,
-      "loss": 0.1398,
+      "loss": 0.3441,
       "step": 1399
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7402356433806478e-05,
-      "loss": 0.1945,
+      "loss": 0.0123,
       "step": 1400
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.740050097411634e-05,
-      "loss": 0.1416,
+      "loss": 0.015,
       "step": 1401
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.73986455144262e-05,
-      "loss": 0.143,
+      "loss": 0.0762,
       "step": 1402
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.739679005473606e-05,
-      "loss": 0.1406,
+      "loss": 0.0202,
       "step": 1403
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7394934595045926e-05,
-      "loss": 0.2462,
+      "loss": 0.0744,
       "step": 1404
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7393079135355787e-05,
-      "loss": 0.1953,
+      "loss": 0.1331,
       "step": 1405
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7391223675665646e-05,
-      "loss": 0.2399,
+      "loss": 0.1316,
       "step": 1406
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7389368215975508e-05,
-      "loss": 0.093,
+      "loss": 0.0792,
       "step": 1407
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7387512756285373e-05,
-      "loss": 0.3519,
+      "loss": 0.1291,
       "step": 1408
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.738565729659523e-05,
-      "loss": 0.187,
+      "loss": 0.1261,
       "step": 1409
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7383801836905093e-05,
-      "loss": 0.1873,
+      "loss": 0.0221,
       "step": 1410
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7381946377214955e-05,
-      "loss": 0.1934,
+      "loss": 0.1223,
       "step": 1411
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.738009091752482e-05,
-      "loss": 0.0899,
+      "loss": 0.0749,
       "step": 1412
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.737823545783468e-05,
-      "loss": 0.185,
+      "loss": 0.0774,
       "step": 1413
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.737637999814454e-05,
-      "loss": 0.0913,
+      "loss": 0.1833,
       "step": 1414
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7374524538454403e-05,
-      "loss": 0.2422,
+      "loss": 0.1217,
       "step": 1415
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7372669078764265e-05,
-      "loss": 0.0941,
+      "loss": 0.0214,
       "step": 1416
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.7370813619074127e-05,
-      "loss": 0.0947,
+      "loss": 0.0693,
       "step": 1417
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.736895815938399e-05,
-      "loss": 0.2761,
+      "loss": 0.1769,
       "step": 1418
     },
     {
       "epoch": 0.39,
       "learning_rate": 1.736710269969385e-05,
-      "loss": 0.0922,
+      "loss": 0.0846,
       "step": 1419
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7365247240003712e-05,
-      "loss": 0.0914,
+      "loss": 0.1838,
       "step": 1420
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7363391780313574e-05,
-      "loss": 0.0911,
+      "loss": 0.0222,
       "step": 1421
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7361536320623436e-05,
-      "loss": 0.0909,
+      "loss": 0.0886,
       "step": 1422
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7359680860933298e-05,
-      "loss": 0.1404,
+      "loss": 0.076,
       "step": 1423
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.735782540124316e-05,
-      "loss": 0.1415,
+      "loss": 0.1536,
       "step": 1424
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.735596994155302e-05,
-      "loss": 0.136,
+      "loss": 0.1267,
       "step": 1425
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7354114481862883e-05,
-      "loss": 0.1944,
+      "loss": 0.0811,
       "step": 1426
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7352259022172745e-05,
-      "loss": 0.0368,
+      "loss": 0.1245,
       "step": 1427
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7350403562482607e-05,
-      "loss": 0.2437,
+      "loss": 0.0875,
       "step": 1428
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.734854810279247e-05,
-      "loss": 0.1408,
+      "loss": 0.1743,
       "step": 1429
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.734669264310233e-05,
-      "loss": 0.2954,
+      "loss": 0.0321,
       "step": 1430
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.734483718341219e-05,
-      "loss": 0.087,
+      "loss": 0.2321,
       "step": 1431
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7342981723722055e-05,
-      "loss": 0.1367,
+      "loss": 0.0706,
       "step": 1432
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7341126264031916e-05,
-      "loss": 0.0337,
+      "loss": 0.0668,
       "step": 1433
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.733927080434178e-05,
-      "loss": 0.0877,
+      "loss": 0.0172,
       "step": 1434
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7337415344651637e-05,
-      "loss": 0.1371,
+      "loss": 0.0145,
       "step": 1435
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7335559884961502e-05,
-      "loss": 0.1941,
+      "loss": 0.0786,
       "step": 1436
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7333704425271364e-05,
-      "loss": 0.1946,
+      "loss": 0.0848,
       "step": 1437
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7331848965581222e-05,
-      "loss": 0.1409,
+      "loss": 0.0699,
       "step": 1438
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7329993505891084e-05,
-      "loss": 0.0329,
+      "loss": 0.0061,
       "step": 1439
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.732813804620095e-05,
-      "loss": 0.3009,
+      "loss": 0.1993,
       "step": 1440
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7326282586510808e-05,
-      "loss": 0.2975,
+      "loss": 0.2292,
       "step": 1441
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.732442712682067e-05,
-      "loss": 0.1403,
+      "loss": 0.0887,
       "step": 1442
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7322571667130532e-05,
-      "loss": 0.0846,
+      "loss": 0.1362,
       "step": 1443
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7320716207440397e-05,
-      "loss": 0.3023,
+      "loss": 0.0697,
       "step": 1444
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7318860747750256e-05,
-      "loss": 0.0335,
+      "loss": 0.2356,
       "step": 1445
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7317005288060117e-05,
-      "loss": 0.0906,
+      "loss": 0.114,
       "step": 1446
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.731514982836998e-05,
-      "loss": 0.2429,
+      "loss": 0.1259,
       "step": 1447
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.731329436867984e-05,
-      "loss": 0.0846,
+      "loss": 0.0949,
       "step": 1448
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7311438908989703e-05,
-      "loss": 0.1381,
+      "loss": 0.0545,
       "step": 1449
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7309583449299565e-05,
-      "loss": 0.087,
+      "loss": 0.2963,
       "step": 1450
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7307727989609427e-05,
-      "loss": 0.1925,
+      "loss": 0.1314,
       "step": 1451
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.730587252991929e-05,
-      "loss": 0.0875,
+      "loss": 0.1744,
       "step": 1452
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.730401707022915e-05,
-      "loss": 0.0863,
+      "loss": 0.1194,
       "step": 1453
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7302161610539012e-05,
-      "loss": 0.2425,
+      "loss": 0.1795,
       "step": 1454
     },
     {
       "epoch": 0.4,
       "learning_rate": 1.7300306150848874e-05,
-      "loss": 0.2433,
+      "loss": 0.0777,
       "step": 1455
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7298450691158736e-05,
-      "loss": 0.089,
+      "loss": 0.0327,
       "step": 1456
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7296595231468598e-05,
-      "loss": 0.2384,
+      "loss": 0.0281,
       "step": 1457
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.729473977177846e-05,
-      "loss": 0.196,
+      "loss": 0.0616,
       "step": 1458
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7292884312088322e-05,
-      "loss": 0.0893,
+      "loss": 0.0855,
       "step": 1459
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7291028852398184e-05,
-      "loss": 0.1391,
+      "loss": 0.1366,
       "step": 1460
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7289173392708045e-05,
-      "loss": 0.1852,
+      "loss": 0.1376,
       "step": 1461
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7287317933017907e-05,
-      "loss": 0.0885,
+      "loss": 0.2657,
       "step": 1462
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.728546247332777e-05,
-      "loss": 0.2477,
+      "loss": 0.0736,
       "step": 1463
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.728360701363763e-05,
-      "loss": 0.2981,
+      "loss": 0.1378,
       "step": 1464
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7281751553947493e-05,
-      "loss": 0.2456,
+      "loss": 0.1323,
       "step": 1465
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7279896094257355e-05,
-      "loss": 0.1384,
+      "loss": 0.0178,
       "step": 1466
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7278040634567213e-05,
-      "loss": 0.1905,
+      "loss": 0.0773,
       "step": 1467
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.727618517487708e-05,
-      "loss": 0.0903,
+      "loss": 0.0774,
       "step": 1468
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.727432971518694e-05,
-      "loss": 0.1367,
+      "loss": 0.0801,
       "step": 1469
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.72724742554968e-05,
-      "loss": 0.141,
+      "loss": 0.2603,
       "step": 1470
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.727061879580666e-05,
-      "loss": 0.0943,
+      "loss": 0.1333,
       "step": 1471
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7268763336116526e-05,
-      "loss": 0.0901,
+      "loss": 0.1205,
       "step": 1472
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7266907876426388e-05,
-      "loss": 0.0873,
+      "loss": 0.1268,
       "step": 1473
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7265052416736246e-05,
-      "loss": 0.1439,
+      "loss": 0.0778,
       "step": 1474
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7263196957046108e-05,
-      "loss": 0.1351,
+      "loss": 0.2231,
       "step": 1475
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7261341497355974e-05,
-      "loss": 0.2409,
+      "loss": 0.0806,
       "step": 1476
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7259486037665832e-05,
-      "loss": 0.0896,
+      "loss": 0.1292,
       "step": 1477
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7257630577975694e-05,
-      "loss": 0.0894,
+      "loss": 0.0859,
       "step": 1478
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7255775118285556e-05,
-      "loss": 0.1406,
+      "loss": 0.0793,
       "step": 1479
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.725391965859542e-05,
-      "loss": 0.0895,
+      "loss": 0.0729,
       "step": 1480
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.725206419890528e-05,
-      "loss": 0.1376,
+      "loss": 0.1153,
       "step": 1481
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.725020873921514e-05,
-      "loss": 0.0866,
+      "loss": 0.0553,
       "step": 1482
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7248353279525003e-05,
-      "loss": 0.0859,
+      "loss": 0.1723,
       "step": 1483
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7246497819834865e-05,
-      "loss": 0.084,
+      "loss": 0.0706,
       "step": 1484
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7244642360144727e-05,
-      "loss": 0.1892,
+      "loss": 0.0155,
       "step": 1485
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.724278690045459e-05,
-      "loss": 0.2398,
+      "loss": 0.0844,
       "step": 1486
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.724093144076445e-05,
-      "loss": 0.0314,
+      "loss": 0.1219,
       "step": 1487
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7239075981074313e-05,
-      "loss": 0.1941,
+      "loss": 0.0733,
       "step": 1488
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7237220521384174e-05,
-      "loss": 0.1957,
+      "loss": 0.0455,
       "step": 1489
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7235365061694036e-05,
-      "loss": 0.1389,
+      "loss": 0.1913,
       "step": 1490
     },
     {
       "epoch": 0.41,
       "learning_rate": 1.7233509602003898e-05,
-      "loss": 0.1438,
+      "loss": 0.0699,
       "step": 1491
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.723165414231376e-05,
-      "loss": 0.0301,
+      "loss": 0.134,
       "step": 1492
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7229798682623622e-05,
-      "loss": 0.2462,
+      "loss": 0.0994,
       "step": 1493
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7227943222933484e-05,
-      "loss": 0.1419,
+      "loss": 0.0218,
       "step": 1494
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7226087763243346e-05,
-      "loss": 0.1944,
+      "loss": 0.2037,
       "step": 1495
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7224232303553208e-05,
-      "loss": 0.3989,
+      "loss": 0.0516,
       "step": 1496
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.722237684386307e-05,
-      "loss": 0.0851,
+      "loss": 0.1364,
       "step": 1497
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.722052138417293e-05,
-      "loss": 0.1363,
+      "loss": 0.1383,
       "step": 1498
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.721866592448279e-05,
-      "loss": 0.1909,
+      "loss": 0.0805,
       "step": 1499
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7216810464792655e-05,
-      "loss": 0.1383,
+      "loss": 0.2012,
       "step": 1500
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7214955005102517e-05,
-      "loss": 0.1937,
+      "loss": 0.1463,
       "step": 1501
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.721309954541238e-05,
-      "loss": 0.1864,
+      "loss": 0.131,
       "step": 1502
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7211244085722237e-05,
-      "loss": 0.0344,
+      "loss": 0.117,
       "step": 1503
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7209388626032103e-05,
-      "loss": 0.1398,
+      "loss": 0.1262,
       "step": 1504
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7207533166341964e-05,
-      "loss": 0.2405,
+      "loss": 0.1563,
       "step": 1505
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7205677706651823e-05,
-      "loss": 0.0848,
+      "loss": 0.1598,
       "step": 1506
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7203822246961685e-05,
-      "loss": 0.2412,
+      "loss": 0.0828,
       "step": 1507
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.720196678727155e-05,
-      "loss": 0.0863,
+      "loss": 0.1923,
       "step": 1508
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7200111327581412e-05,
-      "loss": 0.0849,
+      "loss": 0.0329,
       "step": 1509
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.719825586789127e-05,
-      "loss": 0.1378,
+      "loss": 0.1912,
       "step": 1510
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7196400408201132e-05,
-      "loss": 0.2958,
+      "loss": 0.159,
       "step": 1511
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7194544948510997e-05,
-      "loss": 0.1378,
+      "loss": 0.0306,
       "step": 1512
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7192689488820856e-05,
-      "loss": 0.1883,
+      "loss": 0.167,
       "step": 1513
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7190834029130718e-05,
-      "loss": 0.0364,
+      "loss": 0.1249,
       "step": 1514
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.718897856944058e-05,
-      "loss": 0.1947,
+      "loss": 0.1389,
       "step": 1515
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.718712310975044e-05,
-      "loss": 0.1946,
+      "loss": 0.1479,
       "step": 1516
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7185267650060303e-05,
-      "loss": 0.1943,
+      "loss": 0.1585,
       "step": 1517
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7183412190370165e-05,
-      "loss": 0.0854,
+      "loss": 0.0872,
       "step": 1518
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7181556730680027e-05,
-      "loss": 0.1921,
+      "loss": 0.0929,
       "step": 1519
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.717970127098989e-05,
-      "loss": 0.2367,
+      "loss": 0.1739,
       "step": 1520
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.717784581129975e-05,
-      "loss": 0.2429,
+      "loss": 0.1759,
       "step": 1521
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7175990351609613e-05,
-      "loss": 0.1411,
+      "loss": 0.1127,
       "step": 1522
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7174134891919475e-05,
-      "loss": 0.1823,
+      "loss": 0.1195,
       "step": 1523
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7172279432229337e-05,
-      "loss": 0.1449,
+      "loss": 0.1188,
       "step": 1524
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.71704239725392e-05,
-      "loss": 0.1884,
+      "loss": 0.0309,
       "step": 1525
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.716856851284906e-05,
-      "loss": 0.2344,
+      "loss": 0.235,
       "step": 1526
     },
     {
       "epoch": 0.42,
       "learning_rate": 1.7166713053158922e-05,
-      "loss": 0.1405,
+      "loss": 0.113,
       "step": 1527
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.716485759346878e-05,
-      "loss": 0.2387,
+      "loss": 0.1676,
       "step": 1528
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7163002133778646e-05,
-      "loss": 0.1436,
+      "loss": 0.1119,
       "step": 1529
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7161146674088508e-05,
-      "loss": 0.1884,
+      "loss": 0.0349,
       "step": 1530
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.715929121439837e-05,
-      "loss": 0.1441,
+      "loss": 0.0272,
       "step": 1531
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7157435754708228e-05,
-      "loss": 0.1381,
+      "loss": 0.0782,
       "step": 1532
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7155580295018093e-05,
-      "loss": 0.0455,
+      "loss": 0.2023,
       "step": 1533
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7153724835327955e-05,
-      "loss": 0.1422,
+      "loss": 0.0869,
       "step": 1534
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7151869375637814e-05,
-      "loss": 0.3359,
+      "loss": 0.1046,
       "step": 1535
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7150013915947676e-05,
-      "loss": 0.0928,
+      "loss": 0.0745,
       "step": 1536
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.714815845625754e-05,
-      "loss": 0.1893,
+      "loss": 0.0751,
       "step": 1537
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7146302996567403e-05,
-      "loss": 0.0453,
+      "loss": 0.0227,
       "step": 1538
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.714444753687726e-05,
-      "loss": 0.1807,
+      "loss": 0.1125,
       "step": 1539
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7142592077187123e-05,
-      "loss": 0.0887,
+      "loss": 0.1745,
       "step": 1540
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.714073661749699e-05,
-      "loss": 0.3379,
+      "loss": 0.0794,
       "step": 1541
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7138881157806847e-05,
-      "loss": 0.091,
+      "loss": 0.2564,
       "step": 1542
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.713702569811671e-05,
-      "loss": 0.0876,
+      "loss": 0.1573,
       "step": 1543
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.713517023842657e-05,
-      "loss": 0.1409,
+      "loss": 0.0712,
       "step": 1544
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7133314778736432e-05,
-      "loss": 0.1376,
+      "loss": 0.0429,
       "step": 1545
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7131459319046294e-05,
-      "loss": 0.14,
+      "loss": 0.1135,
       "step": 1546
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7129603859356156e-05,
-      "loss": 0.0376,
+      "loss": 0.1189,
       "step": 1547
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7127748399666018e-05,
-      "loss": 0.1883,
+      "loss": 0.1197,
       "step": 1548
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.712589293997588e-05,
-      "loss": 0.1949,
+      "loss": 0.0715,
       "step": 1549
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7124037480285742e-05,
-      "loss": 0.0905,
+      "loss": 0.0674,
       "step": 1550
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7122182020595604e-05,
-      "loss": 0.0875,
+      "loss": 0.1339,
       "step": 1551
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7120326560905466e-05,
-      "loss": 0.086,
+      "loss": 0.0765,
       "step": 1552
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7118471101215327e-05,
-      "loss": 0.084,
+      "loss": 0.0725,
       "step": 1553
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.711661564152519e-05,
-      "loss": 0.0845,
+      "loss": 0.0495,
       "step": 1554
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.711476018183505e-05,
-      "loss": 0.0314,
+      "loss": 0.073,
       "step": 1555
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7112904722144913e-05,
-      "loss": 0.0831,
+      "loss": 0.1325,
       "step": 1556
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7111049262454775e-05,
-      "loss": 0.1405,
+      "loss": 0.0871,
       "step": 1557
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7109193802764637e-05,
-      "loss": 0.1958,
+      "loss": 0.1689,
       "step": 1558
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.71073383430745e-05,
-      "loss": 0.1923,
+      "loss": 0.1134,
       "step": 1559
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.710548288338436e-05,
-      "loss": 0.1912,
+      "loss": 0.1758,
       "step": 1560
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7103627423694222e-05,
-      "loss": 0.1965,
+      "loss": 0.0909,
       "step": 1561
     },
     {
       "epoch": 0.43,
       "learning_rate": 1.7101771964004084e-05,
-      "loss": 0.1992,
+      "loss": 0.125,
       "step": 1562
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7099916504313946e-05,
-      "loss": 0.0834,
+      "loss": 0.1615,
       "step": 1563
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7098061044623805e-05,
-      "loss": 0.1376,
+      "loss": 0.1198,
       "step": 1564
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.709620558493367e-05,
-      "loss": 0.2546,
+      "loss": 0.0617,
       "step": 1565
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7094350125243532e-05,
-      "loss": 0.0874,
+      "loss": 0.1771,
       "step": 1566
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.709249466555339e-05,
-      "loss": 0.0284,
+      "loss": 0.0412,
       "step": 1567
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7090639205863252e-05,
-      "loss": 0.1944,
+      "loss": 0.071,
       "step": 1568
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7088783746173117e-05,
-      "loss": 0.0289,
+      "loss": 0.1765,
       "step": 1569
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.708692828648298e-05,
-      "loss": 0.087,
+      "loss": 0.0147,
       "step": 1570
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7085072826792838e-05,
-      "loss": 0.2054,
+      "loss": 0.0637,
       "step": 1571
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.70832173671027e-05,
-      "loss": 0.1331,
+      "loss": 0.0249,
       "step": 1572
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7081361907412565e-05,
-      "loss": 0.1371,
+      "loss": 0.1072,
       "step": 1573
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7079506447722423e-05,
-      "loss": 0.14,
+      "loss": 0.1208,
       "step": 1574
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7077650988032285e-05,
-      "loss": 0.0275,
+      "loss": 0.0666,
       "step": 1575
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7075795528342147e-05,
-      "loss": 0.193,
+      "loss": 0.1367,
       "step": 1576
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7073940068652012e-05,
-      "loss": 0.1418,
+      "loss": 0.0524,
       "step": 1577
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.707208460896187e-05,
-      "loss": 0.1996,
+      "loss": 0.1217,
       "step": 1578
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7070229149271733e-05,
-      "loss": 0.1933,
+      "loss": 0.0773,
       "step": 1579
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7068373689581595e-05,
-      "loss": 0.0286,
+      "loss": 0.0629,
       "step": 1580
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7066518229891456e-05,
-      "loss": 0.1443,
+      "loss": 0.0165,
       "step": 1581
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7064662770201318e-05,
-      "loss": 0.0872,
+      "loss": 0.0121,
       "step": 1582
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.706280731051118e-05,
-      "loss": 0.0298,
+      "loss": 0.0754,
       "step": 1583
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7060951850821042e-05,
-      "loss": 0.0283,
+      "loss": 0.0653,
       "step": 1584
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7059096391130904e-05,
-      "loss": 0.1441,
+      "loss": 0.1402,
       "step": 1585
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7057240931440766e-05,
-      "loss": 0.0271,
+      "loss": 0.0672,
       "step": 1586
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7055385471750628e-05,
-      "loss": 0.0834,
+      "loss": 0.0108,
       "step": 1587
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.705353001206049e-05,
-      "loss": 0.0873,
+      "loss": 0.1814,
       "step": 1588
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.705167455237035e-05,
-      "loss": 0.2566,
+      "loss": 0.0195,
       "step": 1589
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7049819092680213e-05,
-      "loss": 0.1923,
+      "loss": 0.1718,
       "step": 1590
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7047963632990075e-05,
-      "loss": 0.309,
+      "loss": 0.3808,
       "step": 1591
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7046108173299937e-05,
-      "loss": 0.1382,
+      "loss": 0.2505,
       "step": 1592
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.70442527136098e-05,
-      "loss": 0.0857,
+      "loss": 0.0744,
       "step": 1593
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.704239725391966e-05,
-      "loss": 0.2504,
+      "loss": 0.2108,
       "step": 1594
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7040541794229523e-05,
-      "loss": 0.14,
+      "loss": 0.1198,
       "step": 1595
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.703868633453938e-05,
-      "loss": 0.1937,
+      "loss": 0.184,
       "step": 1596
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7036830874849246e-05,
-      "loss": 0.1933,
+      "loss": 0.2781,
       "step": 1597
     },
     {
       "epoch": 0.44,
       "learning_rate": 1.7034975415159108e-05,
-      "loss": 0.1889,
+      "loss": 0.1045,
       "step": 1598
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.703311995546897e-05,
-      "loss": 0.1372,
+      "loss": 0.1279,
       "step": 1599
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.703126449577883e-05,
-      "loss": 0.1951,
+      "loss": 0.1205,
       "step": 1600
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.7029409036088694e-05,
-      "loss": 0.0854,
+      "loss": 0.0831,
       "step": 1601
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.7027553576398556e-05,
-      "loss": 0.0874,
+      "loss": 0.1533,
       "step": 1602
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.7025698116708414e-05,
-      "loss": 0.0325,
+      "loss": 0.1031,
       "step": 1603
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.7023842657018276e-05,
-      "loss": 0.0858,
+      "loss": 0.0614,
       "step": 1604
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.702198719732814e-05,
-      "loss": 0.4039,
+      "loss": 0.2293,
       "step": 1605
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.7020131737638003e-05,
-      "loss": 0.0815,
+      "loss": 0.2071,
       "step": 1606
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.701827627794786e-05,
-      "loss": 0.1896,
+      "loss": 0.112,
       "step": 1607
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.7016420818257723e-05,
-      "loss": 0.0852,
+      "loss": 0.0735,
       "step": 1608
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.701456535856759e-05,
-      "loss": 0.1407,
+      "loss": 0.0133,
       "step": 1609
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.7012709898877447e-05,
-      "loss": 0.3039,
+      "loss": 0.099,
       "step": 1610
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.701085443918731e-05,
-      "loss": 0.139,
+      "loss": 0.0747,
       "step": 1611
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.700899897949717e-05,
-      "loss": 0.0338,
+      "loss": 0.1002,
       "step": 1612
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.7007143519807033e-05,
-      "loss": 0.301,
+      "loss": 0.0894,
       "step": 1613
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.7005288060116895e-05,
-      "loss": 0.1954,
+      "loss": 0.0462,
       "step": 1614
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.7003432600426757e-05,
-      "loss": 0.1416,
+      "loss": 0.0935,
       "step": 1615
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.700157714073662e-05,
-      "loss": 0.1365,
+      "loss": 0.0788,
       "step": 1616
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.699972168104648e-05,
-      "loss": 0.3534,
+      "loss": 0.2256,
       "step": 1617
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.6997866221356342e-05,
-      "loss": 0.2426,
+      "loss": 0.0765,
       "step": 1618
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.6996010761666204e-05,
-      "loss": 0.1368,
+      "loss": 0.0263,
       "step": 1619
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.6994155301976066e-05,
-      "loss": 0.1896,
+      "loss": 0.0798,
       "step": 1620
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.6992299842285928e-05,
-      "loss": 0.0417,
+      "loss": 0.0812,
       "step": 1621
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.699044438259579e-05,
-      "loss": 0.2386,
+      "loss": 0.0727,
       "step": 1622
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.698858892290565e-05,
-      "loss": 0.1892,
+      "loss": 0.0212,
       "step": 1623
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.6986733463215513e-05,
-      "loss": 0.1924,
+      "loss": 0.0187,
       "step": 1624
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.6984878003525375e-05,
-      "loss": 0.2422,
+      "loss": 0.1963,
       "step": 1625
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.6983022543835237e-05,
-      "loss": 0.0908,
+      "loss": 0.1267,
       "step": 1626
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.69811670841451e-05,
-      "loss": 0.1411,
+      "loss": 0.1701,
       "step": 1627
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.697931162445496e-05,
-      "loss": 0.1904,
+      "loss": 0.0583,
       "step": 1628
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.6977456164764823e-05,
-      "loss": 0.1821,
+      "loss": 0.0158,
       "step": 1629
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.6975600705074685e-05,
-      "loss": 0.1412,
+      "loss": 0.0148,
       "step": 1630
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.6973745245384547e-05,
-      "loss": 0.0951,
+      "loss": 0.0708,
       "step": 1631
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.6971889785694405e-05,
-      "loss": 0.1874,
+      "loss": 0.2903,
       "step": 1632
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.697003432600427e-05,
-      "loss": 0.1947,
+      "loss": 0.2659,
       "step": 1633
     },
     {
       "epoch": 0.45,
       "learning_rate": 1.6968178866314132e-05,
-      "loss": 0.1835,
+      "loss": 0.0656,
       "step": 1634
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6966323406623994e-05,
-      "loss": 0.1861,
+      "loss": 0.1154,
       "step": 1635
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6964467946933852e-05,
-      "loss": 0.2832,
+      "loss": 0.0684,
       "step": 1636
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6962612487243718e-05,
-      "loss": 0.2415,
+      "loss": 0.0301,
       "step": 1637
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.696075702755358e-05,
-      "loss": 0.3823,
+      "loss": 0.0816,
       "step": 1638
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6958901567863438e-05,
-      "loss": 0.0945,
+      "loss": 0.0589,
       "step": 1639
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.69570461081733e-05,
-      "loss": 0.1926,
+      "loss": 0.2022,
       "step": 1640
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6955190648483165e-05,
-      "loss": 0.0941,
+      "loss": 0.0825,
       "step": 1641
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6953335188793024e-05,
-      "loss": 0.1922,
+      "loss": 0.1332,
       "step": 1642
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6951479729102886e-05,
-      "loss": 0.1015,
+      "loss": 0.0202,
       "step": 1643
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6949624269412747e-05,
-      "loss": 0.1913,
+      "loss": 0.111,
       "step": 1644
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6947768809722613e-05,
-      "loss": 0.1903,
+      "loss": 0.0684,
       "step": 1645
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.694591335003247e-05,
-      "loss": 0.0504,
+      "loss": 0.1361,
       "step": 1646
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6944057890342333e-05,
-      "loss": 0.0982,
+      "loss": 0.1126,
       "step": 1647
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6942202430652195e-05,
-      "loss": 0.1385,
+      "loss": 0.0569,
       "step": 1648
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6940346970962057e-05,
-      "loss": 0.2418,
+      "loss": 0.2469,
       "step": 1649
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.693849151127192e-05,
-      "loss": 0.1398,
+      "loss": 0.1106,
       "step": 1650
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.693663605158178e-05,
-      "loss": 0.1368,
+      "loss": 0.1354,
       "step": 1651
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6934780591891642e-05,
-      "loss": 0.1366,
+      "loss": 0.0744,
       "step": 1652
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6932925132201504e-05,
-      "loss": 0.2905,
+      "loss": 0.0433,
       "step": 1653
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6931069672511366e-05,
-      "loss": 0.0904,
+      "loss": 0.0433,
       "step": 1654
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6929214212821228e-05,
-      "loss": 0.1446,
+      "loss": 0.0465,
       "step": 1655
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.692735875313109e-05,
-      "loss": 0.0381,
+      "loss": 0.0621,
       "step": 1656
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6925503293440952e-05,
-      "loss": 0.2966,
+      "loss": 0.247,
       "step": 1657
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6923647833750814e-05,
-      "loss": 0.2928,
+      "loss": 0.1106,
       "step": 1658
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6921792374060676e-05,
-      "loss": 0.1899,
+      "loss": 0.05,
       "step": 1659
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6919936914370537e-05,
-      "loss": 0.0873,
+      "loss": 0.0611,
       "step": 1660
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.69180814546804e-05,
-      "loss": 0.2425,
+      "loss": 0.0714,
       "step": 1661
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.691622599499026e-05,
-      "loss": 0.0869,
+      "loss": 0.1492,
       "step": 1662
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6914370535300123e-05,
-      "loss": 0.0892,
+      "loss": 0.0218,
       "step": 1663
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6912515075609985e-05,
-      "loss": 0.0367,
+      "loss": 0.0508,
       "step": 1664
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6910659615919847e-05,
-      "loss": 0.0853,
+      "loss": 0.0866,
       "step": 1665
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.690880415622971e-05,
-      "loss": 0.3008,
+      "loss": 0.1415,
       "step": 1666
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.690694869653957e-05,
-      "loss": 0.2509,
+      "loss": 0.0483,
       "step": 1667
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.690509323684943e-05,
-      "loss": 0.1397,
+      "loss": 0.154,
       "step": 1668
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6903237777159294e-05,
-      "loss": 0.1911,
+      "loss": 0.1313,
       "step": 1669
     },
     {
       "epoch": 0.46,
       "learning_rate": 1.6901382317469156e-05,
-      "loss": 0.0875,
+      "loss": 0.0711,
       "step": 1670
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6899526857779015e-05,
-      "loss": 0.1443,
+      "loss": 0.015,
       "step": 1671
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6897671398088876e-05,
-      "loss": 0.0877,
+      "loss": 0.1101,
       "step": 1672
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.689581593839874e-05,
-      "loss": 0.1913,
+      "loss": 0.1635,
       "step": 1673
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6893960478708604e-05,
-      "loss": 0.0855,
+      "loss": 0.1139,
       "step": 1674
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6892105019018462e-05,
-      "loss": 0.1417,
+      "loss": 0.0731,
       "step": 1675
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6890249559328324e-05,
-      "loss": 0.2468,
+      "loss": 0.0348,
       "step": 1676
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6888394099638186e-05,
-      "loss": 0.1897,
+      "loss": 0.0779,
       "step": 1677
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6886538639948048e-05,
-      "loss": 0.0882,
+      "loss": 0.0833,
       "step": 1678
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.688468318025791e-05,
-      "loss": 0.1918,
+      "loss": 0.155,
       "step": 1679
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.688282772056777e-05,
-      "loss": 0.144,
+      "loss": 0.0747,
       "step": 1680
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6880972260877633e-05,
-      "loss": 0.1358,
+      "loss": 0.0734,
       "step": 1681
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6879116801187495e-05,
-      "loss": 0.2925,
+      "loss": 0.0682,
       "step": 1682
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6877261341497357e-05,
-      "loss": 0.0864,
+      "loss": 0.2141,
       "step": 1683
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.687540588180722e-05,
-      "loss": 0.1415,
+      "loss": 0.0171,
       "step": 1684
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.687355042211708e-05,
-      "loss": 0.19,
+      "loss": 0.019,
       "step": 1685
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6871694962426943e-05,
-      "loss": 0.1864,
+      "loss": 0.1167,
       "step": 1686
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6869839502736804e-05,
-      "loss": 0.1946,
+      "loss": 0.117,
       "step": 1687
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6867984043046666e-05,
-      "loss": 0.3883,
+      "loss": 0.0132,
       "step": 1688
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6866128583356528e-05,
-      "loss": 0.1887,
+      "loss": 0.0102,
       "step": 1689
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.686427312366639e-05,
-      "loss": 0.2401,
+      "loss": 0.1586,
       "step": 1690
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6862417663976252e-05,
-      "loss": 0.1402,
+      "loss": 0.2661,
       "step": 1691
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6860562204286114e-05,
-      "loss": 0.1863,
+      "loss": 0.1161,
       "step": 1692
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6858706744595972e-05,
-      "loss": 0.1884,
+      "loss": 0.1455,
       "step": 1693
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6856851284905838e-05,
-      "loss": 0.1908,
+      "loss": 0.1248,
       "step": 1694
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.68549958252157e-05,
-      "loss": 0.1442,
+      "loss": 0.1414,
       "step": 1695
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.685314036552556e-05,
-      "loss": 0.0475,
+      "loss": 0.0821,
       "step": 1696
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.685128490583542e-05,
-      "loss": 0.2772,
+      "loss": 0.0359,
       "step": 1697
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6849429446145285e-05,
-      "loss": 0.1942,
+      "loss": 0.0916,
       "step": 1698
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6847573986455147e-05,
-      "loss": 0.0479,
+      "loss": 0.0462,
       "step": 1699
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6845718526765005e-05,
-      "loss": 0.0939,
+      "loss": 0.135,
       "step": 1700
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6843863067074867e-05,
-      "loss": 0.0945,
+      "loss": 0.0381,
       "step": 1701
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6842007607384733e-05,
-      "loss": 0.0952,
+      "loss": 0.0253,
       "step": 1702
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6840152147694594e-05,
-      "loss": 0.0922,
+      "loss": 0.1866,
       "step": 1703
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6838296688004453e-05,
-      "loss": 0.093,
+      "loss": 0.0665,
       "step": 1704
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.6836441228314315e-05,
-      "loss": 0.2952,
+      "loss": 0.0862,
       "step": 1705
     },
     {
       "epoch": 0.47,
       "learning_rate": 1.683458576862418e-05,
-      "loss": 0.0382,
+      "loss": 0.0148,
       "step": 1706
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.683273030893404e-05,
-      "loss": 0.1387,
+      "loss": 0.207,
       "step": 1707
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.68308748492439e-05,
-      "loss": 0.1937,
+      "loss": 0.0945,
       "step": 1708
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6829019389553762e-05,
-      "loss": 0.0853,
+      "loss": 0.0581,
       "step": 1709
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6827163929863624e-05,
-      "loss": 0.0875,
+      "loss": 0.0146,
       "step": 1710
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6825308470173486e-05,
-      "loss": 0.0899,
+      "loss": 0.1995,
       "step": 1711
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6823453010483348e-05,
-      "loss": 0.1381,
+      "loss": 0.007,
       "step": 1712
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.682159755079321e-05,
-      "loss": 0.0334,
+      "loss": 0.2264,
       "step": 1713
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.681974209110307e-05,
-      "loss": 0.1916,
+      "loss": 0.1293,
       "step": 1714
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6817886631412933e-05,
-      "loss": 0.1402,
+      "loss": 0.0969,
       "step": 1715
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6816031171722795e-05,
-      "loss": 0.198,
+      "loss": 0.107,
       "step": 1716
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6814175712032657e-05,
-      "loss": 0.1428,
+      "loss": 0.0824,
       "step": 1717
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.681232025234252e-05,
-      "loss": 0.194,
+      "loss": 0.1261,
       "step": 1718
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.681046479265238e-05,
-      "loss": 0.0872,
+      "loss": 0.1132,
       "step": 1719
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6808609332962243e-05,
-      "loss": 0.085,
+      "loss": 0.177,
       "step": 1720
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6806753873272105e-05,
-      "loss": 0.1905,
+      "loss": 0.1548,
       "step": 1721
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6804898413581967e-05,
-      "loss": 0.2007,
+      "loss": 0.11,
       "step": 1722
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.680304295389183e-05,
-      "loss": 0.0832,
+      "loss": 0.1957,
       "step": 1723
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.680118749420169e-05,
-      "loss": 0.1365,
+      "loss": 0.0684,
       "step": 1724
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6799332034511552e-05,
-      "loss": 0.1951,
+      "loss": 0.1621,
       "step": 1725
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6797476574821414e-05,
-      "loss": 0.1886,
+      "loss": 0.0582,
       "step": 1726
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6795621115131276e-05,
-      "loss": 0.1893,
+      "loss": 0.0646,
       "step": 1727
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6793765655441138e-05,
-      "loss": 0.1399,
+      "loss": 0.0533,
       "step": 1728
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6791910195750996e-05,
-      "loss": 0.3004,
+      "loss": 0.0684,
       "step": 1729
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.679005473606086e-05,
-      "loss": 0.1407,
+      "loss": 0.2079,
       "step": 1730
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6788199276370723e-05,
-      "loss": 0.137,
+      "loss": 0.0726,
       "step": 1731
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6786343816680585e-05,
-      "loss": 0.2515,
+      "loss": 0.1183,
       "step": 1732
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6784488356990444e-05,
-      "loss": 0.1333,
+      "loss": 0.1027,
       "step": 1733
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.678263289730031e-05,
-      "loss": 0.1402,
+      "loss": 0.0175,
       "step": 1734
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.678077743761017e-05,
-      "loss": 0.1364,
+      "loss": 0.0676,
       "step": 1735
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.677892197792003e-05,
-      "loss": 0.0354,
+      "loss": 0.0355,
       "step": 1736
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.677706651822989e-05,
-      "loss": 0.2419,
+      "loss": 0.0723,
       "step": 1737
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6775211058539757e-05,
-      "loss": 0.1369,
+      "loss": 0.1243,
       "step": 1738
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6773355598849615e-05,
-      "loss": 0.1411,
+      "loss": 0.1034,
       "step": 1739
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6771500139159477e-05,
-      "loss": 0.1358,
+      "loss": 0.1442,
       "step": 1740
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.676964467946934e-05,
-      "loss": 0.182,
+      "loss": 0.0149,
       "step": 1741
     },
     {
       "epoch": 0.48,
       "learning_rate": 1.6767789219779204e-05,
-      "loss": 0.089,
+      "loss": 0.0755,
       "step": 1742
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6765933760089062e-05,
-      "loss": 0.2385,
+      "loss": 0.165,
       "step": 1743
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6764078300398924e-05,
-      "loss": 0.0896,
+      "loss": 0.1027,
       "step": 1744
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6762222840708786e-05,
-      "loss": 0.0393,
+      "loss": 0.0668,
       "step": 1745
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6760367381018648e-05,
-      "loss": 0.1993,
+      "loss": 0.1458,
       "step": 1746
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.675851192132851e-05,
-      "loss": 0.2361,
+      "loss": 0.0607,
       "step": 1747
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6756656461638372e-05,
-      "loss": 0.085,
+      "loss": 0.0199,
       "step": 1748
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6754801001948234e-05,
-      "loss": 0.1373,
+      "loss": 0.0981,
       "step": 1749
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6752945542258096e-05,
-      "loss": 0.0924,
+      "loss": 0.0303,
       "step": 1750
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6751090082567957e-05,
-      "loss": 0.087,
+      "loss": 0.0658,
       "step": 1751
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.674923462287782e-05,
-      "loss": 0.0378,
+      "loss": 0.2779,
       "step": 1752
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.674737916318768e-05,
-      "loss": 0.346,
+      "loss": 0.0544,
       "step": 1753
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6745523703497543e-05,
-      "loss": 0.0901,
+      "loss": 0.268,
       "step": 1754
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6743668243807405e-05,
-      "loss": 0.2484,
+      "loss": 0.1315,
       "step": 1755
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6741812784117267e-05,
-      "loss": 0.1864,
+      "loss": 0.2288,
       "step": 1756
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.673995732442713e-05,
-      "loss": 0.0872,
+      "loss": 0.0593,
       "step": 1757
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.673810186473699e-05,
-      "loss": 0.1369,
+      "loss": 0.1033,
       "step": 1758
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6736246405046852e-05,
-      "loss": 0.2418,
+      "loss": 0.1682,
       "step": 1759
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6734390945356714e-05,
-      "loss": 0.1385,
+      "loss": 0.0373,
       "step": 1760
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6732535485666576e-05,
-      "loss": 0.1881,
+      "loss": 0.1092,
       "step": 1761
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6730680025976438e-05,
-      "loss": 0.0878,
+      "loss": 0.0525,
       "step": 1762
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.67288245662863e-05,
-      "loss": 0.2991,
+      "loss": 0.0598,
       "step": 1763
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6726969106596162e-05,
-      "loss": 0.2919,
+      "loss": 0.1529,
       "step": 1764
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.672511364690602e-05,
-      "loss": 0.0861,
+      "loss": 0.1159,
       "step": 1765
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6723258187215885e-05,
-      "loss": 0.089,
+      "loss": 0.0274,
       "step": 1766
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6721402727525747e-05,
-      "loss": 0.1411,
+      "loss": 0.0228,
       "step": 1767
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6719547267835606e-05,
-      "loss": 0.088,
+      "loss": 0.2266,
       "step": 1768
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6717691808145468e-05,
-      "loss": 0.0879,
+      "loss": 0.0168,
       "step": 1769
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6715836348455333e-05,
-      "loss": 0.141,
+      "loss": 0.075,
       "step": 1770
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6713980888765195e-05,
-      "loss": 0.2372,
+      "loss": 0.0412,
       "step": 1771
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6712125429075053e-05,
-      "loss": 0.1352,
+      "loss": 0.0112,
       "step": 1772
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6710269969384915e-05,
-      "loss": 0.1387,
+      "loss": 0.1155,
       "step": 1773
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.670841450969478e-05,
-      "loss": 0.1386,
+      "loss": 0.0879,
       "step": 1774
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.670655905000464e-05,
-      "loss": 0.0356,
+      "loss": 0.1112,
       "step": 1775
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.67047035903145e-05,
-      "loss": 0.086,
+      "loss": 0.0804,
       "step": 1776
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6702848130624363e-05,
-      "loss": 0.1937,
+      "loss": 0.0101,
       "step": 1777
     },
     {
       "epoch": 0.49,
       "learning_rate": 1.6700992670934228e-05,
-      "loss": 0.292,
+      "loss": 0.0653,
       "step": 1778
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6699137211244086e-05,
-      "loss": 0.0888,
+      "loss": 0.0169,
       "step": 1779
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6697281751553948e-05,
-      "loss": 0.0863,
+      "loss": 0.0092,
       "step": 1780
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.669542629186381e-05,
-      "loss": 0.1371,
+      "loss": 0.0102,
       "step": 1781
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6693570832173672e-05,
-      "loss": 0.1918,
+      "loss": 0.1782,
       "step": 1782
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6691715372483534e-05,
-      "loss": 0.0859,
+      "loss": 0.073,
       "step": 1783
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6689859912793396e-05,
-      "loss": 0.0858,
+      "loss": 0.107,
       "step": 1784
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6688004453103258e-05,
-      "loss": 0.1954,
+      "loss": 0.1425,
       "step": 1785
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.668614899341312e-05,
-      "loss": 0.2434,
+      "loss": 0.0868,
       "step": 1786
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.668429353372298e-05,
-      "loss": 0.0803,
+      "loss": 0.1127,
       "step": 1787
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6682438074032843e-05,
-      "loss": 0.4061,
+      "loss": 0.1075,
       "step": 1788
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6680582614342705e-05,
-      "loss": 0.3033,
+      "loss": 0.0091,
       "step": 1789
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6678727154652567e-05,
-      "loss": 0.0338,
+      "loss": 0.0135,
       "step": 1790
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.667687169496243e-05,
-      "loss": 0.09,
+      "loss": 0.0063,
       "step": 1791
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.667501623527229e-05,
-      "loss": 0.0863,
+      "loss": 0.0069,
       "step": 1792
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6673160775582153e-05,
-      "loss": 0.1941,
+      "loss": 0.0762,
       "step": 1793
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6671305315892014e-05,
-      "loss": 0.1938,
+      "loss": 0.3017,
       "step": 1794
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6669449856201876e-05,
-      "loss": 0.2966,
+      "loss": 0.0841,
       "step": 1795
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6667594396511738e-05,
-      "loss": 0.1928,
+      "loss": 0.071,
       "step": 1796
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6665738936821597e-05,
-      "loss": 0.089,
+      "loss": 0.1114,
       "step": 1797
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6663883477131462e-05,
-      "loss": 0.1899,
+      "loss": 0.2178,
       "step": 1798
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6662028017441324e-05,
-      "loss": 0.141,
+      "loss": 0.2855,
       "step": 1799
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6660172557751186e-05,
-      "loss": 0.1949,
+      "loss": 0.2318,
       "step": 1800
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6658317098061044e-05,
-      "loss": 0.2398,
+      "loss": 0.1224,
       "step": 1801
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.665646163837091e-05,
-      "loss": 0.2938,
+      "loss": 0.0831,
       "step": 1802
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.665460617868077e-05,
-      "loss": 0.0882,
+      "loss": 0.1417,
       "step": 1803
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.665275071899063e-05,
-      "loss": 0.1377,
+      "loss": 0.0829,
       "step": 1804
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.665089525930049e-05,
-      "loss": 0.1377,
+      "loss": 0.0875,
       "step": 1805
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6649039799610357e-05,
-      "loss": 0.1415,
+      "loss": 0.1726,
       "step": 1806
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6647184339920215e-05,
-      "loss": 0.1945,
+      "loss": 0.1008,
       "step": 1807
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6645328880230077e-05,
-      "loss": 0.0939,
+      "loss": 0.0605,
       "step": 1808
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.664347342053994e-05,
-      "loss": 0.1391,
+      "loss": 0.0715,
       "step": 1809
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6641617960849804e-05,
-      "loss": 0.0953,
+      "loss": 0.0304,
       "step": 1810
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6639762501159663e-05,
-      "loss": 0.1359,
+      "loss": 0.0545,
       "step": 1811
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6637907041469525e-05,
-      "loss": 0.0424,
+      "loss": 0.0242,
       "step": 1812
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.6636051581779387e-05,
-      "loss": 0.2397,
+      "loss": 0.1041,
       "step": 1813
     },
     {
       "epoch": 0.5,
       "learning_rate": 1.663419612208925e-05,
-      "loss": 0.1427,
+      "loss": 0.0644,
       "step": 1814
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.663234066239911e-05,
-      "loss": 0.0401,
+      "loss": 0.0782,
       "step": 1815
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6630485202708972e-05,
-      "loss": 0.1886,
+      "loss": 0.1113,
       "step": 1816
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6628629743018834e-05,
-      "loss": 0.2424,
+      "loss": 0.0356,
       "step": 1817
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6626774283328696e-05,
-      "loss": 0.2467,
+      "loss": 0.0669,
       "step": 1818
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6624918823638558e-05,
-      "loss": 0.089,
+      "loss": 0.0105,
       "step": 1819
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.662306336394842e-05,
-      "loss": 0.3515,
+      "loss": 0.1163,
       "step": 1820
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.662120790425828e-05,
-      "loss": 0.1383,
+      "loss": 0.2124,
       "step": 1821
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6619352444568143e-05,
-      "loss": 0.1874,
+      "loss": 0.0647,
       "step": 1822
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6617496984878005e-05,
-      "loss": 0.0902,
+      "loss": 0.0722,
       "step": 1823
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6615641525187867e-05,
-      "loss": 0.0371,
+      "loss": 0.1297,
       "step": 1824
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.661378606549773e-05,
-      "loss": 0.196,
+      "loss": 0.0701,
       "step": 1825
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6611930605807588e-05,
-      "loss": 0.2428,
+      "loss": 0.1005,
       "step": 1826
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6610075146117453e-05,
-      "loss": 0.14,
+      "loss": 0.142,
       "step": 1827
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6608219686427315e-05,
-      "loss": 0.2367,
+      "loss": 0.0149,
       "step": 1828
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6606364226737177e-05,
-      "loss": 0.0376,
+      "loss": 0.0575,
       "step": 1829
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6604508767047035e-05,
-      "loss": 0.09,
+      "loss": 0.0192,
       "step": 1830
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.66026533073569e-05,
-      "loss": 0.1357,
+      "loss": 0.1531,
       "step": 1831
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6600797847666762e-05,
-      "loss": 0.1417,
+      "loss": 0.1139,
       "step": 1832
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.659894238797662e-05,
-      "loss": 0.0915,
+      "loss": 0.0622,
       "step": 1833
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6597086928286483e-05,
-      "loss": 0.0844,
+      "loss": 0.125,
       "step": 1834
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6595231468596348e-05,
-      "loss": 0.2486,
+      "loss": 0.1962,
       "step": 1835
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6593376008906206e-05,
-      "loss": 0.1423,
+      "loss": 0.0695,
       "step": 1836
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6591520549216068e-05,
-      "loss": 0.0884,
+      "loss": 0.1276,
       "step": 1837
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.658966508952593e-05,
-      "loss": 0.087,
+      "loss": 0.2234,
       "step": 1838
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6587809629835795e-05,
-      "loss": 0.1393,
+      "loss": 0.081,
       "step": 1839
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6585954170145654e-05,
-      "loss": 0.086,
+      "loss": 0.1628,
       "step": 1840
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6584098710455516e-05,
-      "loss": 0.0341,
+      "loss": 0.0946,
       "step": 1841
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6582243250765377e-05,
-      "loss": 0.0323,
+      "loss": 0.0752,
       "step": 1842
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.658038779107524e-05,
-      "loss": 0.1891,
+      "loss": 0.1024,
       "step": 1843
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.65785323313851e-05,
-      "loss": 0.0313,
+      "loss": 0.0255,
       "step": 1844
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6576676871694963e-05,
-      "loss": 0.1365,
+      "loss": 0.1836,
       "step": 1845
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6574821412004825e-05,
-      "loss": 0.3657,
+      "loss": 0.1347,
       "step": 1846
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6572965952314687e-05,
-      "loss": 0.1428,
+      "loss": 0.1579,
       "step": 1847
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.657111049262455e-05,
-      "loss": 0.1902,
+      "loss": 0.096,
       "step": 1848
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.656925503293441e-05,
-      "loss": 0.1972,
+      "loss": 0.1352,
       "step": 1849
     },
     {
       "epoch": 0.51,
       "learning_rate": 1.6567399573244272e-05,
-      "loss": 0.2521,
+      "loss": 0.0703,
       "step": 1850
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6565544113554134e-05,
-      "loss": 0.0815,
+      "loss": 0.1906,
       "step": 1851
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6563688653863996e-05,
-      "loss": 0.1939,
+      "loss": 0.0695,
       "step": 1852
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6561833194173858e-05,
-      "loss": 0.1367,
+      "loss": 0.0805,
       "step": 1853
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.655997773448372e-05,
-      "loss": 0.0305,
+      "loss": 0.1162,
       "step": 1854
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6558122274793582e-05,
-      "loss": 0.2512,
+      "loss": 0.0608,
       "step": 1855
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6556266815103444e-05,
-      "loss": 0.142,
+      "loss": 0.0642,
       "step": 1856
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6554411355413306e-05,
-      "loss": 0.1392,
+      "loss": 0.1236,
       "step": 1857
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6552555895723167e-05,
-      "loss": 0.1983,
+      "loss": 0.0182,
       "step": 1858
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.655070043603303e-05,
-      "loss": 0.2473,
+      "loss": 0.0152,
       "step": 1859
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.654884497634289e-05,
-      "loss": 0.1987,
+      "loss": 0.0559,
       "step": 1860
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6546989516652753e-05,
-      "loss": 0.1913,
+      "loss": 0.2346,
       "step": 1861
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.654513405696261e-05,
-      "loss": 0.2411,
+      "loss": 0.014,
       "step": 1862
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6543278597272477e-05,
-      "loss": 0.0362,
+      "loss": 0.01,
       "step": 1863
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.654142313758234e-05,
-      "loss": 0.2885,
+      "loss": 0.0145,
       "step": 1864
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6539567677892197e-05,
-      "loss": 0.0921,
+      "loss": 0.1064,
       "step": 1865
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.653771221820206e-05,
-      "loss": 0.2461,
+      "loss": 0.133,
       "step": 1866
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6535856758511924e-05,
-      "loss": 0.1386,
+      "loss": 0.1115,
       "step": 1867
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6534001298821786e-05,
-      "loss": 0.1424,
+      "loss": 0.158,
       "step": 1868
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6532145839131645e-05,
-      "loss": 0.2419,
+      "loss": 0.0163,
       "step": 1869
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6530290379441506e-05,
-      "loss": 0.0924,
+      "loss": 0.1929,
       "step": 1870
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6528434919751372e-05,
-      "loss": 0.0408,
+      "loss": 0.1034,
       "step": 1871
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.652657946006123e-05,
-      "loss": 0.1915,
+      "loss": 0.1489,
       "step": 1872
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6524724000371092e-05,
-      "loss": 0.1921,
+      "loss": 0.1938,
       "step": 1873
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6522868540680954e-05,
-      "loss": 0.0915,
+      "loss": 0.1606,
       "step": 1874
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.652101308099082e-05,
-      "loss": 0.1437,
+      "loss": 0.1342,
       "step": 1875
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6519157621300678e-05,
-      "loss": 0.1405,
+      "loss": 0.116,
       "step": 1876
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.651730216161054e-05,
-      "loss": 0.2385,
+      "loss": 0.0771,
       "step": 1877
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.65154467019204e-05,
-      "loss": 0.1404,
+      "loss": 0.0419,
       "step": 1878
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6513591242230263e-05,
-      "loss": 0.0886,
+      "loss": 0.0726,
       "step": 1879
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6511735782540125e-05,
-      "loss": 0.2392,
+      "loss": 0.0656,
       "step": 1880
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6509880322849987e-05,
-      "loss": 0.0399,
+      "loss": 0.0332,
       "step": 1881
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.650802486315985e-05,
-      "loss": 0.1919,
+      "loss": 0.125,
       "step": 1882
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.650616940346971e-05,
-      "loss": 0.1419,
+      "loss": 0.0911,
       "step": 1883
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6504313943779573e-05,
-      "loss": 0.0901,
+      "loss": 0.0229,
       "step": 1884
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6502458484089435e-05,
-      "loss": 0.1442,
+      "loss": 0.099,
       "step": 1885
     },
     {
       "epoch": 0.52,
       "learning_rate": 1.6500603024399296e-05,
-      "loss": 0.1375,
+      "loss": 0.1226,
       "step": 1886
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6498747564709158e-05,
-      "loss": 0.1386,
+      "loss": 0.0239,
       "step": 1887
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.649689210501902e-05,
-      "loss": 0.0362,
+      "loss": 0.0631,
       "step": 1888
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6495036645328882e-05,
-      "loss": 0.2427,
+      "loss": 0.0148,
       "step": 1889
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6493181185638744e-05,
-      "loss": 0.191,
+      "loss": 0.0758,
       "step": 1890
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6491325725948606e-05,
-      "loss": 0.188,
+      "loss": 0.1576,
       "step": 1891
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6489470266258468e-05,
-      "loss": 0.3521,
+      "loss": 0.0112,
       "step": 1892
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.648761480656833e-05,
-      "loss": 0.0843,
+      "loss": 0.0142,
       "step": 1893
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6485759346878188e-05,
-      "loss": 0.2379,
+      "loss": 0.0111,
       "step": 1894
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6483903887188053e-05,
-      "loss": 0.1956,
+      "loss": 0.2496,
       "step": 1895
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6482048427497915e-05,
-      "loss": 0.0901,
+      "loss": 0.2374,
       "step": 1896
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6480192967807777e-05,
-      "loss": 0.2381,
+      "loss": 0.2373,
       "step": 1897
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6478337508117635e-05,
-      "loss": 0.0906,
+      "loss": 0.1693,
       "step": 1898
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.64764820484275e-05,
-      "loss": 0.1858,
+      "loss": 0.0519,
       "step": 1899
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6474626588737363e-05,
-      "loss": 0.142,
+      "loss": 0.1931,
       "step": 1900
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.647277112904722e-05,
-      "loss": 0.1373,
+      "loss": 0.0174,
       "step": 1901
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6470915669357083e-05,
-      "loss": 0.2341,
+      "loss": 0.0243,
       "step": 1902
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6469060209666948e-05,
-      "loss": 0.1368,
+      "loss": 0.0311,
       "step": 1903
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.646720474997681e-05,
-      "loss": 0.2967,
+      "loss": 0.0675,
       "step": 1904
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.646534929028667e-05,
-      "loss": 0.1428,
+      "loss": 0.0403,
       "step": 1905
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.646349383059653e-05,
-      "loss": 0.2365,
+      "loss": 0.0795,
       "step": 1906
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6461638370906396e-05,
-      "loss": 0.1398,
+      "loss": 0.151,
       "step": 1907
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6459782911216254e-05,
-      "loss": 0.1889,
+      "loss": 0.0221,
       "step": 1908
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6457927451526116e-05,
-      "loss": 0.0944,
+      "loss": 0.0697,
       "step": 1909
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6456071991835978e-05,
-      "loss": 0.2402,
+      "loss": 0.173,
       "step": 1910
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.645421653214584e-05,
-      "loss": 0.1832,
+      "loss": 0.0931,
       "step": 1911
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.64523610724557e-05,
-      "loss": 0.2323,
+      "loss": 0.1144,
       "step": 1912
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6450505612765564e-05,
-      "loss": 0.3294,
+      "loss": 0.0443,
       "step": 1913
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6448650153075425e-05,
-      "loss": 0.1423,
+      "loss": 0.057,
       "step": 1914
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6446794693385287e-05,
-      "loss": 0.1444,
+      "loss": 0.0724,
       "step": 1915
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.644493923369515e-05,
-      "loss": 0.1412,
+      "loss": 0.022,
       "step": 1916
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.644308377400501e-05,
-      "loss": 0.1894,
+      "loss": 0.1721,
       "step": 1917
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6441228314314873e-05,
-      "loss": 0.1899,
+      "loss": 0.1688,
       "step": 1918
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6439372854624735e-05,
-      "loss": 0.1881,
+      "loss": 0.0709,
       "step": 1919
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.6437517394934597e-05,
-      "loss": 0.0511,
+      "loss": 0.1394,
       "step": 1920
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.643566193524446e-05,
-      "loss": 0.0953,
+      "loss": 0.0196,
       "step": 1921
     },
     {
       "epoch": 0.53,
       "learning_rate": 1.643380647555432e-05,
-      "loss": 0.0928,
+      "loss": 0.0524,
       "step": 1922
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6431951015864182e-05,
-      "loss": 0.1949,
+      "loss": 0.1382,
       "step": 1923
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6430095556174044e-05,
-      "loss": 0.192,
+      "loss": 0.0224,
       "step": 1924
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6428240096483906e-05,
-      "loss": 0.0406,
+      "loss": 0.0494,
       "step": 1925
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6426384636793768e-05,
-      "loss": 0.0372,
+      "loss": 0.0577,
       "step": 1926
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.642452917710363e-05,
-      "loss": 0.0371,
+      "loss": 0.0534,
       "step": 1927
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.642267371741349e-05,
-      "loss": 0.2431,
+      "loss": 0.0705,
       "step": 1928
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6420818257723353e-05,
-      "loss": 0.138,
+      "loss": 0.0767,
       "step": 1929
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6418962798033212e-05,
-      "loss": 0.1934,
+      "loss": 0.0911,
       "step": 1930
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6417107338343077e-05,
-      "loss": 0.032,
+      "loss": 0.099,
       "step": 1931
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.641525187865294e-05,
-      "loss": 0.0315,
+      "loss": 0.1389,
       "step": 1932
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6413396418962798e-05,
-      "loss": 0.1942,
+      "loss": 0.0701,
       "step": 1933
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.641154095927266e-05,
-      "loss": 0.1371,
+      "loss": 0.0117,
       "step": 1934
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6409685499582525e-05,
-      "loss": 0.0846,
+      "loss": 0.1866,
       "step": 1935
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6407830039892387e-05,
-      "loss": 0.0827,
+      "loss": 0.0723,
       "step": 1936
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6405974580202245e-05,
-      "loss": 0.2501,
+      "loss": 0.0362,
       "step": 1937
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6404119120512107e-05,
-      "loss": 0.1433,
+      "loss": 0.0185,
       "step": 1938
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6402263660821972e-05,
-      "loss": 0.3102,
+      "loss": 0.0201,
       "step": 1939
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.640040820113183e-05,
-      "loss": 0.0833,
+      "loss": 0.0123,
       "step": 1940
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6398552741441693e-05,
-      "loss": 0.0318,
+      "loss": 0.0999,
       "step": 1941
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6396697281751554e-05,
-      "loss": 0.1921,
+      "loss": 0.1107,
       "step": 1942
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.639484182206142e-05,
-      "loss": 0.1438,
+      "loss": 0.0085,
       "step": 1943
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6392986362371278e-05,
-      "loss": 0.1351,
+      "loss": 0.0155,
       "step": 1944
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.639113090268114e-05,
-      "loss": 0.1383,
+      "loss": 0.0881,
       "step": 1945
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6389275442991002e-05,
-      "loss": 0.1957,
+      "loss": 0.1401,
       "step": 1946
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6387419983300864e-05,
-      "loss": 0.1418,
+      "loss": 0.0617,
       "step": 1947
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6385564523610726e-05,
-      "loss": 0.3585,
+      "loss": 0.1124,
       "step": 1948
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6383709063920587e-05,
-      "loss": 0.1859,
+      "loss": 0.1,
       "step": 1949
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.638185360423045e-05,
-      "loss": 0.2956,
+      "loss": 0.0163,
       "step": 1950
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.637999814454031e-05,
-      "loss": 0.232,
+      "loss": 0.0196,
       "step": 1951
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6378142684850173e-05,
-      "loss": 0.3169,
+      "loss": 0.0233,
       "step": 1952
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6376287225160035e-05,
-      "loss": 0.1824,
+      "loss": 0.0288,
       "step": 1953
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6374431765469897e-05,
-      "loss": 0.0723,
+      "loss": 0.0239,
       "step": 1954
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.637257630577976e-05,
-      "loss": 0.1019,
+      "loss": 0.1575,
       "step": 1955
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.637072084608962e-05,
-      "loss": 0.099,
+      "loss": 0.0694,
       "step": 1956
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6368865386399482e-05,
-      "loss": 0.2467,
+      "loss": 0.0693,
       "step": 1957
     },
     {
       "epoch": 0.54,
       "learning_rate": 1.6367009926709344e-05,
-      "loss": 0.2394,
+      "loss": 0.0277,
       "step": 1958
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6365154467019206e-05,
-      "loss": 0.1915,
+      "loss": 0.1489,
       "step": 1959
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6363299007329068e-05,
-      "loss": 0.0896,
+      "loss": 0.0676,
       "step": 1960
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.636144354763893e-05,
-      "loss": 0.0373,
+      "loss": 0.011,
       "step": 1961
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.635958808794879e-05,
-      "loss": 0.1949,
+      "loss": 0.1761,
       "step": 1962
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6357732628258654e-05,
-      "loss": 0.0357,
+      "loss": 0.045,
       "step": 1963
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6355877168568516e-05,
-      "loss": 0.2482,
+      "loss": 0.058,
       "step": 1964
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6354021708878377e-05,
-      "loss": 0.2405,
+      "loss": 0.288,
       "step": 1965
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6352166249188236e-05,
-      "loss": 0.0872,
+      "loss": 0.0496,
       "step": 1966
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.63503107894981e-05,
-      "loss": 0.1945,
+      "loss": 0.0963,
       "step": 1967
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6348455329807963e-05,
-      "loss": 0.0869,
+      "loss": 0.0242,
       "step": 1968
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.634659987011782e-05,
-      "loss": 0.0871,
+      "loss": 0.1156,
       "step": 1969
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6344744410427683e-05,
-      "loss": 0.0884,
+      "loss": 0.1606,
       "step": 1970
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.634288895073755e-05,
-      "loss": 0.0857,
+      "loss": 0.0899,
       "step": 1971
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.634103349104741e-05,
-      "loss": 0.2491,
+      "loss": 0.1293,
       "step": 1972
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.633917803135727e-05,
-      "loss": 0.1378,
+      "loss": 0.0191,
       "step": 1973
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.633732257166713e-05,
-      "loss": 0.1367,
+      "loss": 0.1094,
       "step": 1974
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6335467111976993e-05,
-      "loss": 0.1364,
+      "loss": 0.1693,
       "step": 1975
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6333611652286855e-05,
-      "loss": 0.1952,
+      "loss": 0.0712,
       "step": 1976
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6331756192596716e-05,
-      "loss": 0.1445,
+      "loss": 0.1687,
       "step": 1977
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.632990073290658e-05,
-      "loss": 0.0322,
+      "loss": 0.0197,
       "step": 1978
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.632804527321644e-05,
-      "loss": 0.0315,
+      "loss": 0.0763,
       "step": 1979
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6326189813526302e-05,
-      "loss": 0.2496,
+      "loss": 0.1703,
       "step": 1980
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6324334353836164e-05,
-      "loss": 0.251,
+      "loss": 0.0168,
       "step": 1981
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6322478894146026e-05,
-      "loss": 0.1904,
+      "loss": 0.0644,
       "step": 1982
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6320623434455888e-05,
-      "loss": 0.1381,
+      "loss": 0.0778,
       "step": 1983
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.631876797476575e-05,
-      "loss": 0.1467,
+      "loss": 0.1428,
       "step": 1984
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.631691251507561e-05,
-      "loss": 0.1404,
+      "loss": 0.1069,
       "step": 1985
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6315057055385473e-05,
-      "loss": 0.1415,
+      "loss": 0.0229,
       "step": 1986
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6313201595695335e-05,
-      "loss": 0.2445,
+      "loss": 0.0221,
       "step": 1987
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6311346136005197e-05,
-      "loss": 0.1406,
+      "loss": 0.1269,
       "step": 1988
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.630949067631506e-05,
-      "loss": 0.1909,
+      "loss": 0.0158,
       "step": 1989
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.630763521662492e-05,
-      "loss": 0.1394,
+      "loss": 0.0227,
       "step": 1990
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.630577975693478e-05,
-      "loss": 0.1387,
+      "loss": 0.1002,
       "step": 1991
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6303924297244645e-05,
-      "loss": 0.142,
+      "loss": 0.2297,
       "step": 1992
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6302068837554506e-05,
-      "loss": 0.1361,
+      "loss": 0.1048,
       "step": 1993
     },
     {
       "epoch": 0.55,
       "learning_rate": 1.6300213377864368e-05,
-      "loss": 0.1919,
+      "loss": 0.0187,
       "step": 1994
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6298357918174227e-05,
-      "loss": 0.1854,
+      "loss": 0.0185,
       "step": 1995
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6296502458484092e-05,
-      "loss": 0.0881,
+      "loss": 0.0133,
       "step": 1996
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6294646998793954e-05,
-      "loss": 0.1397,
+      "loss": 0.0653,
       "step": 1997
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6292791539103812e-05,
-      "loss": 0.1401,
+      "loss": 0.2524,
       "step": 1998
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6290936079413674e-05,
-      "loss": 0.2401,
+      "loss": 0.0598,
       "step": 1999
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.628908061972354e-05,
-      "loss": 0.2408,
+      "loss": 0.0909,
       "step": 2000
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.62872251600334e-05,
-      "loss": 0.2498,
+      "loss": 0.0689,
       "step": 2001
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.628536970034326e-05,
-      "loss": 0.1944,
+      "loss": 0.0572,
       "step": 2002
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6283514240653122e-05,
-      "loss": 0.0902,
+      "loss": 0.1912,
       "step": 2003
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6281658780962987e-05,
-      "loss": 0.29,
+      "loss": 0.0146,
       "step": 2004
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6279803321272845e-05,
-      "loss": 0.0941,
+      "loss": 0.0572,
       "step": 2005
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6277947861582707e-05,
-      "loss": 0.0969,
+      "loss": 0.0154,
       "step": 2006
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.627609240189257e-05,
-      "loss": 0.14,
+      "loss": 0.0228,
       "step": 2007
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.627423694220243e-05,
-      "loss": 0.0902,
+      "loss": 0.0921,
       "step": 2008
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6272381482512293e-05,
-      "loss": 0.2363,
+      "loss": 0.121,
       "step": 2009
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6270526022822155e-05,
-      "loss": 0.2003,
+      "loss": 0.0227,
       "step": 2010
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6268670563132017e-05,
-      "loss": 0.2354,
+      "loss": 0.1578,
       "step": 2011
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.626681510344188e-05,
-      "loss": 0.1418,
+      "loss": 0.2239,
       "step": 2012
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.626495964375174e-05,
-      "loss": 0.0931,
+      "loss": 0.1686,
       "step": 2013
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6263104184061602e-05,
-      "loss": 0.236,
+      "loss": 0.1176,
       "step": 2014
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6261248724371464e-05,
-      "loss": 0.1395,
+      "loss": 0.1143,
       "step": 2015
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6259393264681326e-05,
-      "loss": 0.0394,
+      "loss": 0.0266,
       "step": 2016
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6257537804991188e-05,
-      "loss": 0.1937,
+      "loss": 0.0685,
       "step": 2017
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.625568234530105e-05,
-      "loss": 0.2377,
+      "loss": 0.0311,
       "step": 2018
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.625382688561091e-05,
-      "loss": 0.0881,
+      "loss": 0.0289,
       "step": 2019
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6251971425920774e-05,
-      "loss": 0.0867,
+      "loss": 0.0285,
       "step": 2020
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6250115966230635e-05,
-      "loss": 0.0894,
+      "loss": 0.1561,
       "step": 2021
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6248260506540497e-05,
-      "loss": 0.1974,
+      "loss": 0.0407,
       "step": 2022
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.624640504685036e-05,
-      "loss": 0.2409,
+      "loss": 0.0866,
       "step": 2023
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.624454958716022e-05,
-      "loss": 0.1877,
+      "loss": 0.0284,
       "step": 2024
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6242694127470083e-05,
-      "loss": 0.2912,
+      "loss": 0.0798,
       "step": 2025
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6240838667779945e-05,
-      "loss": 0.2909,
+      "loss": 0.0712,
       "step": 2026
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6238983208089803e-05,
-      "loss": 0.0858,
+      "loss": 0.0709,
       "step": 2027
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.623712774839967e-05,
-      "loss": 0.1876,
+      "loss": 0.1935,
       "step": 2028
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.623527228870953e-05,
-      "loss": 0.0926,
+      "loss": 0.0844,
       "step": 2029
     },
     {
       "epoch": 0.56,
       "learning_rate": 1.6233416829019392e-05,
-      "loss": 0.1884,
+      "loss": 0.0473,
       "step": 2030
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.623156136932925e-05,
-      "loss": 0.0429,
+      "loss": 0.0931,
       "step": 2031
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6229705909639116e-05,
-      "loss": 0.2399,
+      "loss": 0.0147,
       "step": 2032
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6227850449948978e-05,
-      "loss": 0.1403,
+      "loss": 0.108,
       "step": 2033
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6225994990258836e-05,
-      "loss": 0.1355,
+      "loss": 0.0657,
       "step": 2034
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6224139530568698e-05,
-      "loss": 0.3462,
+      "loss": 0.09,
       "step": 2035
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6222284070878563e-05,
-      "loss": 0.295,
+      "loss": 0.0144,
       "step": 2036
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6220428611188422e-05,
-      "loss": 0.19,
+      "loss": 0.071,
       "step": 2037
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6218573151498284e-05,
-      "loss": 0.0466,
+      "loss": 0.0071,
       "step": 2038
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6216717691808146e-05,
-      "loss": 0.137,
+      "loss": 0.1877,
       "step": 2039
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.621486223211801e-05,
-      "loss": 0.1413,
+      "loss": 0.1643,
       "step": 2040
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.621300677242787e-05,
-      "loss": 0.1453,
+      "loss": 0.0118,
       "step": 2041
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.621115131273773e-05,
-      "loss": 0.1409,
+      "loss": 0.178,
       "step": 2042
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6209295853047593e-05,
-      "loss": 0.1879,
+      "loss": 0.0698,
       "step": 2043
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6207440393357455e-05,
-      "loss": 0.1414,
+      "loss": 0.0181,
       "step": 2044
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6205584933667317e-05,
-      "loss": 0.1882,
+      "loss": 0.1237,
       "step": 2045
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.620372947397718e-05,
-      "loss": 0.1407,
+      "loss": 0.2057,
       "step": 2046
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.620187401428704e-05,
-      "loss": 0.1406,
+      "loss": 0.1094,
       "step": 2047
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6200018554596903e-05,
-      "loss": 0.042,
+      "loss": 0.0182,
       "step": 2048
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6198163094906764e-05,
-      "loss": 0.0859,
+      "loss": 0.1177,
       "step": 2049
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6196307635216626e-05,
-      "loss": 0.1438,
+      "loss": 0.0152,
       "step": 2050
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6194452175526488e-05,
-      "loss": 0.239,
+      "loss": 0.1136,
       "step": 2051
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.619259671583635e-05,
-      "loss": 0.3521,
+      "loss": 0.1504,
       "step": 2052
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6190741256146212e-05,
-      "loss": 0.1917,
+      "loss": 0.0277,
       "step": 2053
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6188885796456074e-05,
-      "loss": 0.1387,
+      "loss": 0.0961,
       "step": 2054
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6187030336765936e-05,
-      "loss": 0.0367,
+      "loss": 0.033,
       "step": 2055
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6185174877075797e-05,
-      "loss": 0.0853,
+      "loss": 0.13,
       "step": 2056
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.618331941738566e-05,
-      "loss": 0.2452,
+      "loss": 0.0201,
       "step": 2057
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.618146395769552e-05,
-      "loss": 0.0881,
+      "loss": 0.0681,
       "step": 2058
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.617960849800538e-05,
-      "loss": 0.1882,
+      "loss": 0.0243,
       "step": 2059
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6177753038315245e-05,
-      "loss": 0.0346,
+      "loss": 0.0326,
       "step": 2060
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6175897578625107e-05,
-      "loss": 0.1367,
+      "loss": 0.0644,
       "step": 2061
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.617404211893497e-05,
-      "loss": 0.1392,
+      "loss": 0.1141,
       "step": 2062
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6172186659244827e-05,
-      "loss": 0.1389,
+      "loss": 0.1011,
       "step": 2063
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6170331199554692e-05,
-      "loss": 0.1363,
+      "loss": 0.1682,
       "step": 2064
     },
     {
       "epoch": 0.57,
       "learning_rate": 1.6168475739864554e-05,
-      "loss": 0.1364,
+      "loss": 0.2075,
       "step": 2065
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6166620280174413e-05,
-      "loss": 0.2992,
+      "loss": 0.0582,
       "step": 2066
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6164764820484275e-05,
-      "loss": 0.1857,
+      "loss": 0.1175,
       "step": 2067
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.616290936079414e-05,
-      "loss": 0.142,
+      "loss": 0.0705,
       "step": 2068
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6161053901104002e-05,
-      "loss": 0.0881,
+      "loss": 0.0979,
       "step": 2069
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.615919844141386e-05,
-      "loss": 0.1409,
+      "loss": 0.1659,
       "step": 2070
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6157342981723722e-05,
-      "loss": 0.3526,
+      "loss": 0.0136,
       "step": 2071
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6155487522033587e-05,
-      "loss": 0.1927,
+      "loss": 0.1259,
       "step": 2072
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6153632062343446e-05,
-      "loss": 0.1382,
+      "loss": 0.1576,
       "step": 2073
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6151776602653308e-05,
-      "loss": 0.0884,
+      "loss": 0.1764,
       "step": 2074
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.614992114296317e-05,
-      "loss": 0.4034,
+      "loss": 0.0726,
       "step": 2075
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.614806568327303e-05,
-      "loss": 0.138,
+      "loss": 0.0639,
       "step": 2076
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6146210223582893e-05,
-      "loss": 0.1884,
+      "loss": 0.1608,
       "step": 2077
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6144354763892755e-05,
-      "loss": 0.1892,
+      "loss": 0.0665,
       "step": 2078
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6142499304202617e-05,
-      "loss": 0.0928,
+      "loss": 0.2021,
       "step": 2079
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.614064384451248e-05,
-      "loss": 0.1919,
+      "loss": 0.1827,
       "step": 2080
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.613878838482234e-05,
-      "loss": 0.2352,
+      "loss": 0.1079,
       "step": 2081
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6136932925132203e-05,
-      "loss": 0.1409,
+      "loss": 0.0952,
       "step": 2082
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6135077465442065e-05,
-      "loss": 0.2866,
+      "loss": 0.0388,
       "step": 2083
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6133222005751926e-05,
-      "loss": 0.3325,
+      "loss": 0.0949,
       "step": 2084
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.613136654606179e-05,
-      "loss": 0.1943,
+      "loss": 0.1057,
       "step": 2085
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.612951108637165e-05,
-      "loss": 0.1446,
+      "loss": 0.06,
       "step": 2086
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6127655626681512e-05,
-      "loss": 0.0556,
+      "loss": 0.0538,
       "step": 2087
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6125800166991374e-05,
-      "loss": 0.0975,
+      "loss": 0.04,
       "step": 2088
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6123944707301236e-05,
-      "loss": 0.1415,
+      "loss": 0.0301,
       "step": 2089
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6122089247611098e-05,
-      "loss": 0.0484,
+      "loss": 0.0617,
       "step": 2090
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.612023378792096e-05,
-      "loss": 0.1374,
+      "loss": 0.0395,
       "step": 2091
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.611837832823082e-05,
-      "loss": 0.1408,
+      "loss": 0.0102,
       "step": 2092
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6116522868540683e-05,
-      "loss": 0.0953,
+      "loss": 0.0999,
       "step": 2093
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6114667408850545e-05,
-      "loss": 0.1924,
+      "loss": 0.008,
       "step": 2094
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6112811949160404e-05,
-      "loss": 0.1354,
+      "loss": 0.1136,
       "step": 2095
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.611095648947027e-05,
-      "loss": 0.1382,
+      "loss": 0.1856,
       "step": 2096
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.610910102978013e-05,
-      "loss": 0.0372,
+      "loss": 0.0663,
       "step": 2097
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6107245570089993e-05,
-      "loss": 0.1363,
+      "loss": 0.1236,
       "step": 2098
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.610539011039985e-05,
-      "loss": 0.3457,
+      "loss": 0.0564,
       "step": 2099
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6103534650709716e-05,
-      "loss": 0.0849,
+      "loss": 0.2329,
       "step": 2100
     },
     {
       "epoch": 0.58,
       "learning_rate": 1.6101679191019578e-05,
-      "loss": 0.0895,
+      "loss": 0.0124,
       "step": 2101
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6099823731329437e-05,
-      "loss": 0.296,
+      "loss": 0.0693,
       "step": 2102
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.60979682716393e-05,
-      "loss": 0.1935,
+      "loss": 0.1356,
       "step": 2103
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6096112811949164e-05,
-      "loss": 0.0871,
+      "loss": 0.1272,
       "step": 2104
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6094257352259022e-05,
-      "loss": 0.249,
+      "loss": 0.0943,
       "step": 2105
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6092401892568884e-05,
-      "loss": 0.305,
+      "loss": 0.1199,
       "step": 2106
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6090546432878746e-05,
-      "loss": 0.0867,
+      "loss": 0.0286,
       "step": 2107
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.608869097318861e-05,
-      "loss": 0.2402,
+      "loss": 0.1624,
       "step": 2108
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.608683551349847e-05,
-      "loss": 0.141,
+      "loss": 0.1644,
       "step": 2109
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6084980053808332e-05,
-      "loss": 0.0893,
+      "loss": 0.0942,
       "step": 2110
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6083124594118194e-05,
-      "loss": 0.1391,
+      "loss": 0.0888,
       "step": 2111
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6081269134428055e-05,
-      "loss": 0.1862,
+      "loss": 0.1175,
       "step": 2112
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6079413674737917e-05,
-      "loss": 0.0891,
+      "loss": 0.1741,
       "step": 2113
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.607755821504778e-05,
-      "loss": 0.1409,
+      "loss": 0.144,
       "step": 2114
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.607570275535764e-05,
-      "loss": 0.0358,
+      "loss": 0.0649,
       "step": 2115
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6073847295667503e-05,
-      "loss": 0.0909,
+      "loss": 0.0506,
       "step": 2116
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6071991835977365e-05,
-      "loss": 0.0878,
+      "loss": 0.0903,
       "step": 2117
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6070136376287227e-05,
-      "loss": 0.1933,
+      "loss": 0.1056,
       "step": 2118
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.606828091659709e-05,
-      "loss": 0.1941,
+      "loss": 0.1003,
       "step": 2119
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.606642545690695e-05,
-      "loss": 0.1366,
+      "loss": 0.0643,
       "step": 2120
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6064569997216812e-05,
-      "loss": 0.2468,
+      "loss": 0.1163,
       "step": 2121
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6062714537526674e-05,
-      "loss": 0.2463,
+      "loss": 0.1079,
       "step": 2122
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6060859077836536e-05,
-      "loss": 0.085,
+      "loss": 0.1647,
       "step": 2123
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6059003618146398e-05,
-      "loss": 0.2438,
+      "loss": 0.1843,
       "step": 2124
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.605714815845626e-05,
-      "loss": 0.1419,
+      "loss": 0.0236,
       "step": 2125
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.605529269876612e-05,
-      "loss": 0.1855,
+      "loss": 0.0302,
       "step": 2126
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6053437239075984e-05,
-      "loss": 0.1382,
+      "loss": 0.0959,
       "step": 2127
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6051581779385842e-05,
-      "loss": 0.1403,
+      "loss": 0.1944,
       "step": 2128
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6049726319695707e-05,
-      "loss": 0.2413,
+      "loss": 0.2036,
       "step": 2129
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.604787086000557e-05,
-      "loss": 0.1378,
+      "loss": 0.1184,
       "step": 2130
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6046015400315428e-05,
-      "loss": 0.0872,
+      "loss": 0.0219,
       "step": 2131
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.604415994062529e-05,
-      "loss": 0.1397,
+      "loss": 0.033,
       "step": 2132
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6042304480935155e-05,
-      "loss": 0.2409,
+      "loss": 0.1018,
       "step": 2133
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6040449021245013e-05,
-      "loss": 0.196,
+      "loss": 0.0386,
       "step": 2134
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6038593561554875e-05,
-      "loss": 0.0859,
+      "loss": 0.1556,
       "step": 2135
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6036738101864737e-05,
-      "loss": 0.0903,
+      "loss": 0.0233,
       "step": 2136
     },
     {
       "epoch": 0.59,
       "learning_rate": 1.6034882642174602e-05,
-      "loss": 0.2447,
+      "loss": 0.0732,
       "step": 2137
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.603302718248446e-05,
-      "loss": 0.0906,
+      "loss": 0.027,
       "step": 2138
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6031171722794323e-05,
-      "loss": 0.1423,
+      "loss": 0.0221,
       "step": 2139
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6029316263104184e-05,
-      "loss": 0.0922,
+      "loss": 0.0188,
       "step": 2140
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6027460803414046e-05,
-      "loss": 0.0893,
+      "loss": 0.0182,
       "step": 2141
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6025605343723908e-05,
-      "loss": 0.242,
+      "loss": 0.1365,
       "step": 2142
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.602374988403377e-05,
-      "loss": 0.189,
+      "loss": 0.0683,
       "step": 2143
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6021894424343632e-05,
-      "loss": 0.0885,
+      "loss": 0.0131,
       "step": 2144
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6020038964653494e-05,
-      "loss": 0.1395,
+      "loss": 0.0725,
       "step": 2145
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6018183504963356e-05,
-      "loss": 0.1391,
+      "loss": 0.0107,
       "step": 2146
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6016328045273218e-05,
-      "loss": 0.1375,
+      "loss": 0.0553,
       "step": 2147
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.601447258558308e-05,
-      "loss": 0.1411,
+      "loss": 0.1076,
       "step": 2148
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.601261712589294e-05,
-      "loss": 0.3924,
+      "loss": 0.0774,
       "step": 2149
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6010761666202803e-05,
-      "loss": 0.1404,
+      "loss": 0.0597,
       "step": 2150
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6008906206512665e-05,
-      "loss": 0.1899,
+      "loss": 0.1732,
       "step": 2151
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6007050746822527e-05,
-      "loss": 0.1425,
+      "loss": 0.0207,
       "step": 2152
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.600519528713239e-05,
-      "loss": 0.2409,
+      "loss": 0.1226,
       "step": 2153
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.600333982744225e-05,
-      "loss": 0.0886,
+      "loss": 0.0097,
       "step": 2154
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.6001484367752112e-05,
-      "loss": 0.0377,
+      "loss": 0.1114,
       "step": 2155
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.599962890806197e-05,
-      "loss": 0.1884,
+      "loss": 0.0809,
       "step": 2156
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.5997773448371836e-05,
-      "loss": 0.1911,
+      "loss": 0.2268,
       "step": 2157
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.5995917988681698e-05,
-      "loss": 0.1362,
+      "loss": 0.0147,
       "step": 2158
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.599406252899156e-05,
-      "loss": 0.1882,
+      "loss": 0.0187,
       "step": 2159
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.599220706930142e-05,
-      "loss": 0.0856,
+      "loss": 0.1957,
       "step": 2160
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.5990351609611284e-05,
-      "loss": 0.192,
+      "loss": 0.1376,
       "step": 2161
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.5988496149921146e-05,
-      "loss": 0.1907,
+      "loss": 0.0224,
       "step": 2162
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.5986640690231004e-05,
-      "loss": 0.0898,
+      "loss": 0.062,
       "step": 2163
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.5984785230540866e-05,
-      "loss": 0.1417,
+      "loss": 0.1228,
       "step": 2164
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.598292977085073e-05,
-      "loss": 0.144,
+      "loss": 0.102,
       "step": 2165
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.5981074311160593e-05,
-      "loss": 0.2965,
+      "loss": 0.0708,
       "step": 2166
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.597921885147045e-05,
-      "loss": 0.1942,
+      "loss": 0.0255,
       "step": 2167
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.5977363391780313e-05,
-      "loss": 0.1929,
+      "loss": 0.1094,
       "step": 2168
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.597550793209018e-05,
-      "loss": 0.1887,
+      "loss": 0.0245,
       "step": 2169
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.5973652472400037e-05,
-      "loss": 0.234,
+      "loss": 0.0566,
       "step": 2170
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.59717970127099e-05,
-      "loss": 0.1919,
+      "loss": 0.0765,
       "step": 2171
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.596994155301976e-05,
-      "loss": 0.1405,
+      "loss": 0.1036,
       "step": 2172
     },
     {
       "epoch": 0.6,
       "learning_rate": 1.5968086093329623e-05,
-      "loss": 0.1403,
+      "loss": 0.2447,
       "step": 2173
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5966230633639485e-05,
-      "loss": 0.1425,
+      "loss": 0.0803,
       "step": 2174
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5964375173949347e-05,
-      "loss": 0.0948,
+      "loss": 0.0654,
       "step": 2175
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.596251971425921e-05,
-      "loss": 0.2401,
+      "loss": 0.1233,
       "step": 2176
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.596066425456907e-05,
-      "loss": 0.1393,
+      "loss": 0.0731,
       "step": 2177
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5958808794878932e-05,
-      "loss": 0.2406,
+      "loss": 0.0169,
       "step": 2178
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5956953335188794e-05,
-      "loss": 0.1915,
+      "loss": 0.0993,
       "step": 2179
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5955097875498656e-05,
-      "loss": 0.1916,
+      "loss": 0.1234,
       "step": 2180
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5953242415808518e-05,
-      "loss": 0.2405,
+      "loss": 0.1528,
       "step": 2181
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.595138695611838e-05,
-      "loss": 0.1412,
+      "loss": 0.1676,
       "step": 2182
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.594953149642824e-05,
-      "loss": 0.1421,
+      "loss": 0.1143,
       "step": 2183
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5947676036738103e-05,
-      "loss": 0.0928,
+      "loss": 0.071,
       "step": 2184
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5945820577047965e-05,
-      "loss": 0.1442,
+      "loss": 0.1136,
       "step": 2185
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5943965117357827e-05,
-      "loss": 0.146,
+      "loss": 0.1015,
       "step": 2186
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.594210965766769e-05,
-      "loss": 0.187,
+      "loss": 0.1137,
       "step": 2187
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.594025419797755e-05,
-      "loss": 0.291,
+      "loss": 0.0686,
       "step": 2188
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5938398738287413e-05,
-      "loss": 0.087,
+      "loss": 0.043,
       "step": 2189
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5936543278597275e-05,
-      "loss": 0.1441,
+      "loss": 0.0206,
       "step": 2190
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5934687818907136e-05,
-      "loss": 0.2454,
+      "loss": 0.0761,
       "step": 2191
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5932832359216995e-05,
-      "loss": 0.1396,
+      "loss": 0.1757,
       "step": 2192
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.593097689952686e-05,
-      "loss": 0.0414,
+      "loss": 0.0679,
       "step": 2193
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5929121439836722e-05,
-      "loss": 0.0411,
+      "loss": 0.0587,
       "step": 2194
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5927265980146584e-05,
-      "loss": 0.0888,
+      "loss": 0.0145,
       "step": 2195
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5925410520456442e-05,
-      "loss": 0.0889,
+      "loss": 0.0824,
       "step": 2196
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5923555060766308e-05,
-      "loss": 0.34,
+      "loss": 0.1091,
       "step": 2197
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.592169960107617e-05,
-      "loss": 0.143,
+      "loss": 0.1418,
       "step": 2198
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5919844141386028e-05,
-      "loss": 0.1943,
+      "loss": 0.0229,
       "step": 2199
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.591798868169589e-05,
-      "loss": 0.3493,
+      "loss": 0.0589,
       "step": 2200
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5916133222005755e-05,
-      "loss": 0.1327,
+      "loss": 0.0216,
       "step": 2201
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5914277762315614e-05,
-      "loss": 0.1392,
+      "loss": 0.058,
       "step": 2202
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5912422302625476e-05,
-      "loss": 0.1419,
+      "loss": 0.0935,
       "step": 2203
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5910566842935337e-05,
-      "loss": 0.0899,
+      "loss": 0.0121,
       "step": 2204
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5908711383245203e-05,
-      "loss": 0.0912,
+      "loss": 0.0515,
       "step": 2205
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.590685592355506e-05,
-      "loss": 0.1359,
+      "loss": 0.0501,
       "step": 2206
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5905000463864923e-05,
-      "loss": 0.1908,
+      "loss": 0.0719,
       "step": 2207
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5903145004174785e-05,
-      "loss": 0.1396,
+      "loss": 0.0999,
       "step": 2208
     },
     {
       "epoch": 0.61,
       "learning_rate": 1.5901289544484647e-05,
-      "loss": 0.1841,
+      "loss": 0.2282,
       "step": 2209
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.589943408479451e-05,
-      "loss": 0.1377,
+      "loss": 0.0433,
       "step": 2210
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.589757862510437e-05,
-      "loss": 0.2491,
+      "loss": 0.0872,
       "step": 2211
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5895723165414232e-05,
-      "loss": 0.2404,
+      "loss": 0.1042,
       "step": 2212
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5893867705724094e-05,
-      "loss": 0.1411,
+      "loss": 0.0192,
       "step": 2213
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5892012246033956e-05,
-      "loss": 0.0365,
+      "loss": 0.063,
       "step": 2214
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5890156786343818e-05,
-      "loss": 0.0873,
+      "loss": 0.1896,
       "step": 2215
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.588830132665368e-05,
-      "loss": 0.3002,
+      "loss": 0.0279,
       "step": 2216
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.588644586696354e-05,
-      "loss": 0.0863,
+      "loss": 0.1139,
       "step": 2217
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5884590407273404e-05,
-      "loss": 0.0865,
+      "loss": 0.0974,
       "step": 2218
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5882734947583265e-05,
-      "loss": 0.2421,
+      "loss": 0.0602,
       "step": 2219
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5880879487893127e-05,
-      "loss": 0.0899,
+      "loss": 0.0707,
       "step": 2220
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.587902402820299e-05,
-      "loss": 0.1827,
+      "loss": 0.0319,
       "step": 2221
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.587716856851285e-05,
-      "loss": 0.1363,
+      "loss": 0.218,
       "step": 2222
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5875313108822713e-05,
-      "loss": 0.1905,
+      "loss": 0.2005,
       "step": 2223
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5873457649132575e-05,
-      "loss": 0.0367,
+      "loss": 0.1027,
       "step": 2224
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5871602189442437e-05,
-      "loss": 0.1959,
+      "loss": 0.1373,
       "step": 2225
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.58697467297523e-05,
-      "loss": 0.0881,
+      "loss": 0.042,
       "step": 2226
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.586789127006216e-05,
-      "loss": 0.0892,
+      "loss": 0.0698,
       "step": 2227
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.586603581037202e-05,
-      "loss": 0.1401,
+      "loss": 0.1313,
       "step": 2228
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5864180350681884e-05,
-      "loss": 0.3402,
+      "loss": 0.0615,
       "step": 2229
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5862324890991746e-05,
-      "loss": 0.1368,
+      "loss": 0.2223,
       "step": 2230
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5860469431301604e-05,
-      "loss": 0.3478,
+      "loss": 0.0534,
       "step": 2231
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5858613971611466e-05,
-      "loss": 0.1967,
+      "loss": 0.0239,
       "step": 2232
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.585675851192133e-05,
-      "loss": 0.1403,
+      "loss": 0.0235,
       "step": 2233
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5854903052231193e-05,
-      "loss": 0.2468,
+      "loss": 0.0666,
       "step": 2234
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5853047592541052e-05,
-      "loss": 0.0371,
+      "loss": 0.1583,
       "step": 2235
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5851192132850914e-05,
-      "loss": 0.1867,
+      "loss": 0.1181,
       "step": 2236
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.584933667316078e-05,
-      "loss": 0.2416,
+      "loss": 0.2333,
       "step": 2237
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5847481213470638e-05,
-      "loss": 0.1393,
+      "loss": 0.069,
       "step": 2238
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.58456257537805e-05,
-      "loss": 0.1403,
+      "loss": 0.0672,
       "step": 2239
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.584377029409036e-05,
-      "loss": 0.0913,
+      "loss": 0.028,
       "step": 2240
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5841914834400227e-05,
-      "loss": 0.1935,
+      "loss": 0.0876,
       "step": 2241
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5840059374710085e-05,
-      "loss": 0.1385,
+      "loss": 0.0652,
       "step": 2242
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.5838203915019947e-05,
-      "loss": 0.1398,
+      "loss": 0.0323,
       "step": 2243
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.583634845532981e-05,
-      "loss": 0.1399,
+      "loss": 0.0703,
       "step": 2244
     },
     {
       "epoch": 0.62,
       "learning_rate": 1.583449299563967e-05,
-      "loss": 0.1372,
+      "loss": 0.0276,
       "step": 2245
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5832637535949533e-05,
-      "loss": 0.0911,
+      "loss": 0.119,
       "step": 2246
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5830782076259394e-05,
-      "loss": 0.241,
+      "loss": 0.2008,
       "step": 2247
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5828926616569256e-05,
-      "loss": 0.0894,
+      "loss": 0.1013,
       "step": 2248
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5827071156879118e-05,
-      "loss": 0.1429,
+      "loss": 0.0626,
       "step": 2249
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.582521569718898e-05,
-      "loss": 0.1985,
+      "loss": 0.0542,
       "step": 2250
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5823360237498842e-05,
-      "loss": 0.0389,
+      "loss": 0.0931,
       "step": 2251
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5821504777808704e-05,
-      "loss": 0.1941,
+      "loss": 0.0153,
       "step": 2252
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5819649318118566e-05,
-      "loss": 0.1935,
+      "loss": 0.0526,
       "step": 2253
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5817793858428428e-05,
-      "loss": 0.091,
+      "loss": 0.1136,
       "step": 2254
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.581593839873829e-05,
-      "loss": 0.4449,
+      "loss": 0.1512,
       "step": 2255
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.581408293904815e-05,
-      "loss": 0.0369,
+      "loss": 0.1501,
       "step": 2256
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5812227479358013e-05,
-      "loss": 0.1925,
+      "loss": 0.1075,
       "step": 2257
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5810372019667875e-05,
-      "loss": 0.3396,
+      "loss": 0.0416,
       "step": 2258
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5808516559977737e-05,
-      "loss": 0.0886,
+      "loss": 0.1297,
       "step": 2259
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5806661100287595e-05,
-      "loss": 0.1929,
+      "loss": 0.0208,
       "step": 2260
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.580480564059746e-05,
-      "loss": 0.4031,
+      "loss": 0.1083,
       "step": 2261
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5802950180907322e-05,
-      "loss": 0.0893,
+      "loss": 0.0331,
       "step": 2262
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5801094721217184e-05,
-      "loss": 0.1396,
+      "loss": 0.0884,
       "step": 2263
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5799239261527043e-05,
-      "loss": 0.0894,
+      "loss": 0.0702,
       "step": 2264
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5797383801836908e-05,
-      "loss": 0.1419,
+      "loss": 0.0625,
       "step": 2265
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.579552834214677e-05,
-      "loss": 0.3489,
+      "loss": 0.1416,
       "step": 2266
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.579367288245663e-05,
-      "loss": 0.041,
+      "loss": 0.0311,
       "step": 2267
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.579181742276649e-05,
-      "loss": 0.145,
+      "loss": 0.1409,
       "step": 2268
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5789961963076356e-05,
-      "loss": 0.0394,
+      "loss": 0.0655,
       "step": 2269
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5788106503386217e-05,
-      "loss": 0.0397,
+      "loss": 0.075,
       "step": 2270
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5786251043696076e-05,
-      "loss": 0.294,
+      "loss": 0.0834,
       "step": 2271
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5784395584005938e-05,
-      "loss": 0.1928,
+      "loss": 0.029,
       "step": 2272
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5782540124315803e-05,
-      "loss": 0.3389,
+      "loss": 0.1039,
       "step": 2273
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.578068466462566e-05,
-      "loss": 0.2417,
+      "loss": 0.116,
       "step": 2274
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5778829204935523e-05,
-      "loss": 0.1423,
+      "loss": 0.0965,
       "step": 2275
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5776973745245385e-05,
-      "loss": 0.1998,
+      "loss": 0.1349,
       "step": 2276
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5775118285555247e-05,
-      "loss": 0.1418,
+      "loss": 0.1126,
       "step": 2277
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.577326282586511e-05,
-      "loss": 0.0905,
+      "loss": 0.2243,
       "step": 2278
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.577140736617497e-05,
-      "loss": 0.2344,
+      "loss": 0.0993,
       "step": 2279
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5769551906484833e-05,
-      "loss": 0.0419,
+      "loss": 0.0316,
       "step": 2280
     },
     {
       "epoch": 0.63,
       "learning_rate": 1.5767696446794695e-05,
-      "loss": 0.0938,
+      "loss": 0.1027,
       "step": 2281
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5765840987104557e-05,
-      "loss": 0.2416,
+      "loss": 0.0916,
       "step": 2282
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.576398552741442e-05,
-      "loss": 0.0912,
+      "loss": 0.064,
       "step": 2283
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.576213006772428e-05,
-      "loss": 0.1829,
+      "loss": 0.1029,
       "step": 2284
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5760274608034142e-05,
-      "loss": 0.2917,
+      "loss": 0.019,
       "step": 2285
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5758419148344004e-05,
-      "loss": 0.1909,
+      "loss": 0.0512,
       "step": 2286
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5756563688653866e-05,
-      "loss": 0.2916,
+      "loss": 0.0621,
       "step": 2287
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5754708228963728e-05,
-      "loss": 0.1911,
+      "loss": 0.1429,
       "step": 2288
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5752852769273586e-05,
-      "loss": 0.0914,
+      "loss": 0.0885,
       "step": 2289
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.575099730958345e-05,
-      "loss": 0.1404,
+      "loss": 0.1969,
       "step": 2290
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5749141849893313e-05,
-      "loss": 0.138,
+      "loss": 0.1058,
       "step": 2291
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5747286390203175e-05,
-      "loss": 0.0913,
+      "loss": 0.0681,
       "step": 2292
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5745430930513034e-05,
-      "loss": 0.1929,
+      "loss": 0.0157,
       "step": 2293
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.57435754708229e-05,
-      "loss": 0.2875,
+      "loss": 0.0161,
       "step": 2294
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.574172001113276e-05,
-      "loss": 0.2368,
+      "loss": 0.0317,
       "step": 2295
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.573986455144262e-05,
-      "loss": 0.2879,
+      "loss": 0.0911,
       "step": 2296
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.573800909175248e-05,
-      "loss": 0.0929,
+      "loss": 0.1432,
       "step": 2297
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5736153632062346e-05,
-      "loss": 0.2363,
+      "loss": 0.0245,
       "step": 2298
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5734298172372205e-05,
-      "loss": 0.0938,
+      "loss": 0.0453,
       "step": 2299
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5732442712682067e-05,
-      "loss": 0.0943,
+      "loss": 0.1639,
       "step": 2300
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.573058725299193e-05,
-      "loss": 0.2429,
+      "loss": 0.1566,
       "step": 2301
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5728731793301794e-05,
-      "loss": 0.1919,
+      "loss": 0.1503,
       "step": 2302
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5726876333611652e-05,
-      "loss": 0.1456,
+      "loss": 0.0209,
       "step": 2303
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5725020873921514e-05,
-      "loss": 0.1909,
+      "loss": 0.0118,
       "step": 2304
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5723165414231376e-05,
-      "loss": 0.0461,
+      "loss": 0.0546,
       "step": 2305
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5721309954541238e-05,
-      "loss": 0.1398,
+      "loss": 0.2187,
       "step": 2306
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.57194544948511e-05,
-      "loss": 0.0953,
+      "loss": 0.0621,
       "step": 2307
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5717599035160962e-05,
-      "loss": 0.0914,
+      "loss": 0.0214,
       "step": 2308
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5715743575470824e-05,
-      "loss": 0.2321,
+      "loss": 0.1131,
       "step": 2309
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5713888115780685e-05,
-      "loss": 0.2386,
+      "loss": 0.062,
       "step": 2310
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5712032656090547e-05,
-      "loss": 0.0394,
+      "loss": 0.1087,
       "step": 2311
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.571017719640041e-05,
-      "loss": 0.0894,
+      "loss": 0.1127,
       "step": 2312
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.570832173671027e-05,
-      "loss": 0.1393,
+      "loss": 0.0895,
       "step": 2313
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5706466277020133e-05,
-      "loss": 0.0854,
+      "loss": 0.1147,
       "step": 2314
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5704610817329995e-05,
-      "loss": 0.0355,
+      "loss": 0.1447,
       "step": 2315
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.5702755357639857e-05,
-      "loss": 0.035,
+      "loss": 0.0777,
       "step": 2316
     },
     {
       "epoch": 0.64,
       "learning_rate": 1.570089989794972e-05,
-      "loss": 0.1914,
+      "loss": 0.0818,
       "step": 2317
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.569904443825958e-05,
-      "loss": 0.1408,
+      "loss": 0.0205,
       "step": 2318
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5697188978569442e-05,
-      "loss": 0.1431,
+      "loss": 0.0886,
       "step": 2319
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5695333518879304e-05,
-      "loss": 0.2471,
+      "loss": 0.0947,
       "step": 2320
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5693478059189166e-05,
-      "loss": 0.0853,
+      "loss": 0.022,
       "step": 2321
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5691622599499028e-05,
-      "loss": 0.0289,
+      "loss": 0.0229,
       "step": 2322
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.568976713980889e-05,
-      "loss": 0.3077,
+      "loss": 0.0728,
       "step": 2323
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.568791168011875e-05,
-      "loss": 0.1965,
+      "loss": 0.1904,
       "step": 2324
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.568605622042861e-05,
-      "loss": 0.3012,
+      "loss": 0.0226,
       "step": 2325
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5684200760738475e-05,
-      "loss": 0.2431,
+      "loss": 0.1065,
       "step": 2326
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5682345301048337e-05,
-      "loss": 0.1968,
+      "loss": 0.02,
       "step": 2327
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5680489841358196e-05,
-      "loss": 0.0301,
+      "loss": 0.1142,
       "step": 2328
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5678634381668058e-05,
-      "loss": 0.1425,
+      "loss": 0.0934,
       "step": 2329
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5676778921977923e-05,
-      "loss": 0.1417,
+      "loss": 0.0905,
       "step": 2330
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5674923462287785e-05,
-      "loss": 0.0873,
+      "loss": 0.0193,
       "step": 2331
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5673068002597643e-05,
-      "loss": 0.1927,
+      "loss": 0.2152,
       "step": 2332
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5671212542907505e-05,
-      "loss": 0.3004,
+      "loss": 0.1053,
       "step": 2333
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.566935708321737e-05,
-      "loss": 0.3554,
+      "loss": 0.1221,
       "step": 2334
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.566750162352723e-05,
-      "loss": 0.19,
+      "loss": 0.0208,
       "step": 2335
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.566564616383709e-05,
-      "loss": 0.0858,
+      "loss": 0.2155,
       "step": 2336
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5663790704146953e-05,
-      "loss": 0.2969,
+      "loss": 0.0214,
       "step": 2337
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5661935244456818e-05,
-      "loss": 0.2468,
+      "loss": 0.1162,
       "step": 2338
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5660079784766676e-05,
-      "loss": 0.2957,
+      "loss": 0.1,
       "step": 2339
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5658224325076538e-05,
-      "loss": 0.0402,
+      "loss": 0.1974,
       "step": 2340
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.56563688653864e-05,
-      "loss": 0.2375,
+      "loss": 0.1162,
       "step": 2341
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5654513405696262e-05,
-      "loss": 0.1918,
+      "loss": 0.1163,
       "step": 2342
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5652657946006124e-05,
-      "loss": 0.1377,
+      "loss": 0.0646,
       "step": 2343
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5650802486315986e-05,
-      "loss": 0.0936,
+      "loss": 0.147,
       "step": 2344
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5648947026625848e-05,
-      "loss": 0.09,
+      "loss": 0.2244,
       "step": 2345
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.564709156693571e-05,
-      "loss": 0.2381,
+      "loss": 0.1616,
       "step": 2346
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.564523610724557e-05,
-      "loss": 0.0994,
+      "loss": 0.0711,
       "step": 2347
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5643380647555433e-05,
-      "loss": 0.1873,
+      "loss": 0.0644,
       "step": 2348
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5641525187865295e-05,
-      "loss": 0.096,
+      "loss": 0.125,
       "step": 2349
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5639669728175157e-05,
-      "loss": 0.0954,
+      "loss": 0.2036,
       "step": 2350
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.563781426848502e-05,
-      "loss": 0.1482,
+      "loss": 0.0345,
       "step": 2351
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.563595880879488e-05,
-      "loss": 0.2398,
+      "loss": 0.0312,
       "step": 2352
     },
     {
       "epoch": 0.65,
       "learning_rate": 1.5634103349104743e-05,
-      "loss": 0.1399,
+      "loss": 0.1491,
       "step": 2353
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5632247889414604e-05,
-      "loss": 0.2918,
+      "loss": 0.0556,
       "step": 2354
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5630392429724466e-05,
-      "loss": 0.1417,
+      "loss": 0.0458,
       "step": 2355
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5628536970034328e-05,
-      "loss": 0.0424,
+      "loss": 0.022,
       "step": 2356
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5626681510344187e-05,
-      "loss": 0.24,
+      "loss": 0.0993,
       "step": 2357
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5624826050654052e-05,
-      "loss": 0.0907,
+      "loss": 0.0902,
       "step": 2358
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5622970590963914e-05,
-      "loss": 0.0408,
+      "loss": 0.1613,
       "step": 2359
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5621115131273776e-05,
-      "loss": 0.0945,
+      "loss": 0.2031,
       "step": 2360
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5619259671583634e-05,
-      "loss": 0.1354,
+      "loss": 0.0361,
       "step": 2361
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.56174042118935e-05,
-      "loss": 0.1922,
+      "loss": 0.1009,
       "step": 2362
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.561554875220336e-05,
-      "loss": 0.1408,
+      "loss": 0.1567,
       "step": 2363
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.561369329251322e-05,
-      "loss": 0.4029,
+      "loss": 0.0566,
       "step": 2364
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.561183783282308e-05,
-      "loss": 0.0353,
+      "loss": 0.0372,
       "step": 2365
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5609982373132947e-05,
-      "loss": 0.1425,
+      "loss": 0.1025,
       "step": 2366
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.560812691344281e-05,
-      "loss": 0.185,
+      "loss": 0.0951,
       "step": 2367
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5606271453752667e-05,
-      "loss": 0.1409,
+      "loss": 0.0199,
       "step": 2368
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.560441599406253e-05,
-      "loss": 0.1849,
+      "loss": 0.1437,
       "step": 2369
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5602560534372394e-05,
-      "loss": 0.1446,
+      "loss": 0.0155,
       "step": 2370
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5600705074682253e-05,
-      "loss": 0.1926,
+      "loss": 0.2177,
       "step": 2371
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5598849614992115e-05,
-      "loss": 0.0872,
+      "loss": 0.0459,
       "step": 2372
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5596994155301977e-05,
-      "loss": 0.0362,
+      "loss": 0.0624,
       "step": 2373
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.559513869561184e-05,
-      "loss": 0.0342,
+      "loss": 0.0608,
       "step": 2374
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.55932832359217e-05,
-      "loss": 0.2485,
+      "loss": 0.0676,
       "step": 2375
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5591427776231562e-05,
-      "loss": 0.1421,
+      "loss": 0.0283,
       "step": 2376
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5589572316541424e-05,
-      "loss": 0.1363,
+      "loss": 0.0191,
       "step": 2377
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5587716856851286e-05,
-      "loss": 0.1958,
+      "loss": 0.0719,
       "step": 2378
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5585861397161148e-05,
-      "loss": 0.1896,
+      "loss": 0.0641,
       "step": 2379
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.558400593747101e-05,
-      "loss": 0.1934,
+      "loss": 0.1773,
       "step": 2380
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.558215047778087e-05,
-      "loss": 0.1894,
+      "loss": 0.171,
       "step": 2381
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5580295018090733e-05,
-      "loss": 0.1974,
+      "loss": 0.0654,
       "step": 2382
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5578439558400595e-05,
-      "loss": 0.1923,
+      "loss": 0.1658,
       "step": 2383
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5576584098710457e-05,
-      "loss": 0.1371,
+      "loss": 0.0958,
       "step": 2384
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.557472863902032e-05,
-      "loss": 0.1949,
+      "loss": 0.0958,
       "step": 2385
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.557287317933018e-05,
-      "loss": 0.1394,
+      "loss": 0.0945,
       "step": 2386
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5571017719640043e-05,
-      "loss": 0.0867,
+      "loss": 0.1671,
       "step": 2387
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5569162259949905e-05,
-      "loss": 0.1356,
+      "loss": 0.0201,
       "step": 2388
     },
     {
       "epoch": 0.66,
       "learning_rate": 1.5567306800259766e-05,
-      "loss": 0.0864,
+      "loss": 0.0736,
       "step": 2389
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.556545134056963e-05,
-      "loss": 0.196,
+      "loss": 0.0296,
       "step": 2390
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.556359588087949e-05,
-      "loss": 0.2956,
+      "loss": 0.0805,
       "step": 2391
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5561740421189352e-05,
-      "loss": 0.0354,
+      "loss": 0.1048,
       "step": 2392
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.555988496149921e-05,
-      "loss": 0.1397,
+      "loss": 0.0667,
       "step": 2393
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5558029501809076e-05,
-      "loss": 0.1395,
+      "loss": 0.0635,
       "step": 2394
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5556174042118938e-05,
-      "loss": 0.1384,
+      "loss": 0.0365,
       "step": 2395
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5554318582428796e-05,
-      "loss": 0.1406,
+      "loss": 0.0895,
       "step": 2396
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5552463122738658e-05,
-      "loss": 0.0871,
+      "loss": 0.0253,
       "step": 2397
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5550607663048523e-05,
-      "loss": 0.1955,
+      "loss": 0.1393,
       "step": 2398
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5548752203358385e-05,
-      "loss": 0.1389,
+      "loss": 0.1599,
       "step": 2399
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5546896743668244e-05,
-      "loss": 0.0882,
+      "loss": 0.0288,
       "step": 2400
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5545041283978106e-05,
-      "loss": 0.1407,
+      "loss": 0.1678,
       "step": 2401
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.554318582428797e-05,
-      "loss": 0.19,
+      "loss": 0.32,
       "step": 2402
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.554133036459783e-05,
-      "loss": 0.2413,
+      "loss": 0.1526,
       "step": 2403
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.553947490490769e-05,
-      "loss": 0.1402,
+      "loss": 0.2034,
       "step": 2404
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5537619445217553e-05,
-      "loss": 0.1362,
+      "loss": 0.0167,
       "step": 2405
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5535763985527418e-05,
-      "loss": 0.2946,
+      "loss": 0.1222,
       "step": 2406
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5533908525837277e-05,
-      "loss": 0.1393,
+      "loss": 0.0621,
       "step": 2407
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.553205306614714e-05,
-      "loss": 0.2394,
+      "loss": 0.0366,
       "step": 2408
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5530197606457e-05,
-      "loss": 0.0892,
+      "loss": 0.0614,
       "step": 2409
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5528342146766862e-05,
-      "loss": 0.1424,
+      "loss": 0.0917,
       "step": 2410
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5526486687076724e-05,
-      "loss": 0.3945,
+      "loss": 0.0631,
       "step": 2411
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5524631227386586e-05,
-      "loss": 0.1408,
+      "loss": 0.1503,
       "step": 2412
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5522775767696448e-05,
-      "loss": 0.1946,
+      "loss": 0.0285,
       "step": 2413
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.552092030800631e-05,
-      "loss": 0.1397,
+      "loss": 0.084,
       "step": 2414
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5519064848316172e-05,
-      "loss": 0.138,
+      "loss": 0.0504,
       "step": 2415
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5517209388626034e-05,
-      "loss": 0.0911,
+      "loss": 0.0662,
       "step": 2416
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5515353928935895e-05,
-      "loss": 0.187,
+      "loss": 0.0265,
       "step": 2417
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5513498469245757e-05,
-      "loss": 0.1485,
+      "loss": 0.0653,
       "step": 2418
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.551164300955562e-05,
-      "loss": 0.1416,
+      "loss": 0.0608,
       "step": 2419
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.550978754986548e-05,
-      "loss": 0.1913,
+      "loss": 0.1191,
       "step": 2420
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5507932090175343e-05,
-      "loss": 0.2442,
+      "loss": 0.0652,
       "step": 2421
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5506076630485205e-05,
-      "loss": 0.1908,
+      "loss": 0.1339,
       "step": 2422
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5504221170795067e-05,
-      "loss": 0.0939,
+      "loss": 0.1726,
       "step": 2423
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.550236571110493e-05,
-      "loss": 0.0403,
+      "loss": 0.0649,
       "step": 2424
     },
     {
       "epoch": 0.67,
       "learning_rate": 1.5500510251414787e-05,
-      "loss": 0.0906,
+      "loss": 0.1177,
       "step": 2425
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5498654791724652e-05,
-      "loss": 0.1399,
+      "loss": 0.0661,
       "step": 2426
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5496799332034514e-05,
-      "loss": 0.1409,
+      "loss": 0.0593,
       "step": 2427
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5494943872344376e-05,
-      "loss": 0.1421,
+      "loss": 0.1539,
       "step": 2428
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5493088412654235e-05,
-      "loss": 0.1929,
+      "loss": 0.0205,
       "step": 2429
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5491232952964096e-05,
-      "loss": 0.1415,
+      "loss": 0.0968,
       "step": 2430
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.548937749327396e-05,
-      "loss": 0.1412,
+      "loss": 0.0668,
       "step": 2431
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.548752203358382e-05,
-      "loss": 0.1951,
+      "loss": 0.1566,
       "step": 2432
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5485666573893682e-05,
-      "loss": 0.0907,
+      "loss": 0.0654,
       "step": 2433
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5483811114203544e-05,
-      "loss": 0.0887,
+      "loss": 0.0697,
       "step": 2434
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.548195565451341e-05,
-      "loss": 0.1383,
+      "loss": 0.1695,
       "step": 2435
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5480100194823268e-05,
-      "loss": 0.1372,
+      "loss": 0.0575,
       "step": 2436
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.547824473513313e-05,
-      "loss": 0.2489,
+      "loss": 0.0302,
       "step": 2437
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.547638927544299e-05,
-      "loss": 0.0881,
+      "loss": 0.189,
       "step": 2438
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5474533815752853e-05,
-      "loss": 0.0884,
+      "loss": 0.0562,
       "step": 2439
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5472678356062715e-05,
-      "loss": 0.1924,
+      "loss": 0.1246,
       "step": 2440
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5470822896372577e-05,
-      "loss": 0.1443,
+      "loss": 0.2075,
       "step": 2441
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.546896743668244e-05,
-      "loss": 0.1914,
+      "loss": 0.1105,
       "step": 2442
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.54671119769923e-05,
-      "loss": 0.0322,
+      "loss": 0.1281,
       "step": 2443
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5465256517302163e-05,
-      "loss": 0.0836,
+      "loss": 0.0209,
       "step": 2444
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5463401057612024e-05,
-      "loss": 0.0845,
+      "loss": 0.0685,
       "step": 2445
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5461545597921886e-05,
-      "loss": 0.138,
+      "loss": 0.1196,
       "step": 2446
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5459690138231748e-05,
-      "loss": 0.1368,
+      "loss": 0.0688,
       "step": 2447
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.545783467854161e-05,
-      "loss": 0.2507,
+      "loss": 0.1526,
       "step": 2448
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5455979218851472e-05,
-      "loss": 0.1855,
+      "loss": 0.0796,
       "step": 2449
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5454123759161334e-05,
-      "loss": 0.188,
+      "loss": 0.0325,
       "step": 2450
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5452268299471196e-05,
-      "loss": 0.2498,
+      "loss": 0.1298,
       "step": 2451
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5450412839781058e-05,
-      "loss": 0.3578,
+      "loss": 0.0901,
       "step": 2452
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.544855738009092e-05,
-      "loss": 0.0296,
+      "loss": 0.0238,
       "step": 2453
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5446701920400778e-05,
-      "loss": 0.2434,
+      "loss": 0.1033,
       "step": 2454
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5444846460710643e-05,
-      "loss": 0.1391,
+      "loss": 0.1219,
       "step": 2455
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5442991001020505e-05,
-      "loss": 0.0338,
+      "loss": 0.0696,
       "step": 2456
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5441135541330367e-05,
-      "loss": 0.3997,
+      "loss": 0.2103,
       "step": 2457
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5439280081640225e-05,
-      "loss": 0.085,
+      "loss": 0.0146,
       "step": 2458
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.543742462195009e-05,
-      "loss": 0.1396,
+      "loss": 0.0893,
       "step": 2459
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.5435569162259953e-05,
-      "loss": 0.0875,
+      "loss": 0.0223,
       "step": 2460
     },
     {
       "epoch": 0.68,
       "learning_rate": 1.543371370256981e-05,
-      "loss": 0.1421,
+      "loss": 0.1427,
       "step": 2461
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5431858242879673e-05,
-      "loss": 0.3469,
+      "loss": 0.0192,
       "step": 2462
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5430002783189538e-05,
-      "loss": 0.1845,
+      "loss": 0.0362,
       "step": 2463
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.54281473234994e-05,
-      "loss": 0.0872,
+      "loss": 0.2569,
       "step": 2464
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.542629186380926e-05,
-      "loss": 0.0894,
+      "loss": 0.0168,
       "step": 2465
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.542443640411912e-05,
-      "loss": 0.0911,
+      "loss": 0.1707,
       "step": 2466
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5422580944428986e-05,
-      "loss": 0.0903,
+      "loss": 0.0362,
       "step": 2467
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5420725484738844e-05,
-      "loss": 0.1879,
+      "loss": 0.1124,
       "step": 2468
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5418870025048706e-05,
-      "loss": 0.0858,
+      "loss": 0.1752,
       "step": 2469
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5417014565358568e-05,
-      "loss": 0.2395,
+      "loss": 0.0895,
       "step": 2470
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.541515910566843e-05,
-      "loss": 0.1375,
+      "loss": 0.0425,
       "step": 2471
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.541330364597829e-05,
-      "loss": 0.1938,
+      "loss": 0.0759,
       "step": 2472
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5411448186288153e-05,
-      "loss": 0.192,
+      "loss": 0.049,
       "step": 2473
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5409592726598015e-05,
-      "loss": 0.1413,
+      "loss": 0.0811,
       "step": 2474
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5407737266907877e-05,
-      "loss": 0.1403,
+      "loss": 0.1843,
       "step": 2475
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.540588180721774e-05,
-      "loss": 0.1863,
+      "loss": 0.1219,
       "step": 2476
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.54040263475276e-05,
-      "loss": 0.037,
+      "loss": 0.0296,
       "step": 2477
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5402170887837463e-05,
-      "loss": 0.0904,
+      "loss": 0.0638,
       "step": 2478
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5400315428147325e-05,
-      "loss": 0.2459,
+      "loss": 0.0829,
       "step": 2479
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5398459968457187e-05,
-      "loss": 0.1378,
+      "loss": 0.0187,
       "step": 2480
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.539660450876705e-05,
-      "loss": 0.1918,
+      "loss": 0.0134,
       "step": 2481
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.539474904907691e-05,
-      "loss": 0.1416,
+      "loss": 0.2184,
       "step": 2482
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5392893589386772e-05,
-      "loss": 0.2395,
+      "loss": 0.1717,
       "step": 2483
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5391038129696634e-05,
-      "loss": 0.1845,
+      "loss": 0.0117,
       "step": 2484
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5389182670006496e-05,
-      "loss": 0.1389,
+      "loss": 0.2155,
       "step": 2485
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5387327210316358e-05,
-      "loss": 0.0871,
+      "loss": 0.0168,
       "step": 2486
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.538547175062622e-05,
-      "loss": 0.2428,
+      "loss": 0.0153,
       "step": 2487
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.538361629093608e-05,
-      "loss": 0.2908,
+      "loss": 0.0649,
       "step": 2488
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5381760831245943e-05,
-      "loss": 0.0901,
+      "loss": 0.0455,
       "step": 2489
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5379905371555802e-05,
-      "loss": 0.1355,
+      "loss": 0.1578,
       "step": 2490
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5378049911865667e-05,
-      "loss": 0.1348,
+      "loss": 0.0435,
       "step": 2491
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.537619445217553e-05,
-      "loss": 0.0877,
+      "loss": 0.1545,
       "step": 2492
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.537433899248539e-05,
-      "loss": 0.0887,
+      "loss": 0.1227,
       "step": 2493
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.537248353279525e-05,
-      "loss": 0.091,
+      "loss": 0.0165,
       "step": 2494
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5370628073105115e-05,
-      "loss": 0.0888,
+      "loss": 0.0162,
       "step": 2495
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5368772613414976e-05,
-      "loss": 0.0928,
+      "loss": 0.0715,
       "step": 2496
     },
     {
       "epoch": 0.69,
       "learning_rate": 1.5366917153724835e-05,
-      "loss": 0.0914,
+      "loss": 0.0661,
       "step": 2497
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5365061694034697e-05,
-      "loss": 0.1914,
+      "loss": 0.0679,
       "step": 2498
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5363206234344562e-05,
-      "loss": 0.0352,
+      "loss": 0.0664,
       "step": 2499
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.536135077465442e-05,
-      "loss": 0.3416,
+      "loss": 0.0808,
       "step": 2500
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5359495314964282e-05,
-      "loss": 0.0884,
+      "loss": 0.013,
       "step": 2501
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5357639855274144e-05,
-      "loss": 0.1911,
+      "loss": 0.0109,
       "step": 2502
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.535578439558401e-05,
-      "loss": 0.1373,
+      "loss": 0.0804,
       "step": 2503
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5353928935893868e-05,
-      "loss": 0.1384,
+      "loss": 0.1499,
       "step": 2504
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.535207347620373e-05,
-      "loss": 0.0841,
+      "loss": 0.2043,
       "step": 2505
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5350218016513592e-05,
-      "loss": 0.3513,
+      "loss": 0.0661,
       "step": 2506
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5348362556823454e-05,
-      "loss": 0.1388,
+      "loss": 0.0289,
       "step": 2507
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5346507097133316e-05,
-      "loss": 0.244,
+      "loss": 0.0193,
       "step": 2508
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5344651637443177e-05,
-      "loss": 0.1353,
+      "loss": 0.1222,
       "step": 2509
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.534279617775304e-05,
-      "loss": 0.1869,
+      "loss": 0.0184,
       "step": 2510
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.53409407180629e-05,
-      "loss": 0.035,
+      "loss": 0.0669,
       "step": 2511
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5339085258372763e-05,
-      "loss": 0.3048,
+      "loss": 0.1499,
       "step": 2512
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5337229798682625e-05,
-      "loss": 0.2415,
+      "loss": 0.1253,
       "step": 2513
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5335374338992487e-05,
-      "loss": 0.2432,
+      "loss": 0.0639,
       "step": 2514
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.533351887930235e-05,
-      "loss": 0.1386,
+      "loss": 0.143,
       "step": 2515
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.533166341961221e-05,
-      "loss": 0.2396,
+      "loss": 0.1032,
       "step": 2516
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5329807959922072e-05,
-      "loss": 0.2447,
+      "loss": 0.022,
       "step": 2517
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5327952500231934e-05,
-      "loss": 0.1437,
+      "loss": 0.0617,
       "step": 2518
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5326097040541796e-05,
-      "loss": 0.0399,
+      "loss": 0.0151,
       "step": 2519
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5324241580851658e-05,
-      "loss": 0.1362,
+      "loss": 0.0606,
       "step": 2520
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.532238612116152e-05,
-      "loss": 0.142,
+      "loss": 0.1775,
       "step": 2521
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.532053066147138e-05,
-      "loss": 0.2389,
+      "loss": 0.1435,
       "step": 2522
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5318675201781244e-05,
-      "loss": 0.1928,
+      "loss": 0.0238,
       "step": 2523
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5316819742091105e-05,
-      "loss": 0.2874,
+      "loss": 0.228,
       "step": 2524
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5314964282400967e-05,
-      "loss": 0.2355,
+      "loss": 0.0939,
       "step": 2525
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5313108822710826e-05,
-      "loss": 0.2883,
+      "loss": 0.2054,
       "step": 2526
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.531125336302069e-05,
-      "loss": 0.1431,
+      "loss": 0.1005,
       "step": 2527
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5309397903330553e-05,
-      "loss": 0.1439,
+      "loss": 0.0864,
       "step": 2528
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.530754244364041e-05,
-      "loss": 0.0969,
+      "loss": 0.06,
       "step": 2529
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.5305686983950273e-05,
-      "loss": 0.0469,
+      "loss": 0.0235,
       "step": 2530
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.530383152426014e-05,
-      "loss": 0.1914,
+      "loss": 0.0929,
       "step": 2531
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.530197606457e-05,
-      "loss": 0.1438,
+      "loss": 0.0217,
       "step": 2532
     },
     {
       "epoch": 0.7,
       "learning_rate": 1.530012060487986e-05,
-      "loss": 0.1446,
+      "loss": 0.1289,
       "step": 2533
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.529826514518972e-05,
-      "loss": 0.1934,
+      "loss": 0.0713,
       "step": 2534
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5296409685499586e-05,
-      "loss": 0.2405,
+      "loss": 0.1153,
       "step": 2535
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5294554225809445e-05,
-      "loss": 0.0923,
+      "loss": 0.144,
       "step": 2536
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5292698766119306e-05,
-      "loss": 0.0912,
+      "loss": 0.1666,
       "step": 2537
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5290843306429168e-05,
-      "loss": 0.2393,
+      "loss": 0.0894,
       "step": 2538
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.528898784673903e-05,
-      "loss": 0.0917,
+      "loss": 0.3268,
       "step": 2539
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5287132387048892e-05,
-      "loss": 0.1441,
+      "loss": 0.1502,
       "step": 2540
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5285276927358754e-05,
-      "loss": 0.2449,
+      "loss": 0.0776,
       "step": 2541
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5283421467668616e-05,
-      "loss": 0.1452,
+      "loss": 0.1015,
       "step": 2542
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5281566007978478e-05,
-      "loss": 0.1402,
+      "loss": 0.0285,
       "step": 2543
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.527971054828834e-05,
-      "loss": 0.14,
+      "loss": 0.0368,
       "step": 2544
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.52778550885982e-05,
-      "loss": 0.1429,
+      "loss": 0.112,
       "step": 2545
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5275999628908063e-05,
-      "loss": 0.0903,
+      "loss": 0.1741,
       "step": 2546
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5274144169217925e-05,
-      "loss": 0.0918,
+      "loss": 0.0711,
       "step": 2547
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5272288709527787e-05,
-      "loss": 0.1361,
+      "loss": 0.0841,
       "step": 2548
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.527043324983765e-05,
-      "loss": 0.1882,
+      "loss": 0.1753,
       "step": 2549
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.526857779014751e-05,
-      "loss": 0.3377,
+      "loss": 0.0239,
       "step": 2550
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5266722330457373e-05,
-      "loss": 0.29,
+      "loss": 0.1951,
       "step": 2551
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5264866870767234e-05,
-      "loss": 0.1938,
+      "loss": 0.0821,
       "step": 2552
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5263011411077096e-05,
-      "loss": 0.04,
+      "loss": 0.0776,
       "step": 2553
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5261155951386958e-05,
-      "loss": 0.2401,
+      "loss": 0.2117,
       "step": 2554
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.525930049169682e-05,
-      "loss": 0.1917,
+      "loss": 0.0844,
       "step": 2555
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5257445032006682e-05,
-      "loss": 0.1905,
+      "loss": 0.158,
       "step": 2556
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5255589572316542e-05,
-      "loss": 0.0385,
+      "loss": 0.136,
       "step": 2557
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5253734112626404e-05,
-      "loss": 0.3958,
+      "loss": 0.0221,
       "step": 2558
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5251878652936268e-05,
-      "loss": 0.1365,
+      "loss": 0.0617,
       "step": 2559
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5250023193246128e-05,
-      "loss": 0.0396,
+      "loss": 0.0336,
       "step": 2560
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.524816773355599e-05,
-      "loss": 0.0396,
+      "loss": 0.1787,
       "step": 2561
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5246312273865851e-05,
-      "loss": 0.0398,
+      "loss": 0.1771,
       "step": 2562
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5244456814175715e-05,
-      "loss": 0.0891,
+      "loss": 0.1563,
       "step": 2563
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5242601354485575e-05,
-      "loss": 0.1416,
+      "loss": 0.0598,
       "step": 2564
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5240745894795437e-05,
-      "loss": 0.0889,
+      "loss": 0.1261,
       "step": 2565
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5238890435105297e-05,
-      "loss": 0.1384,
+      "loss": 0.0677,
       "step": 2566
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.523703497541516e-05,
-      "loss": 0.0884,
+      "loss": 0.0342,
       "step": 2567
     },
     {
       "epoch": 0.71,
       "learning_rate": 1.5235179515725023e-05,
-      "loss": 0.1934,
+      "loss": 0.0759,
       "step": 2568
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5233324056034885e-05,
-      "loss": 0.2489,
+      "loss": 0.0244,
       "step": 2569
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5231468596344745e-05,
-      "loss": 0.0338,
+      "loss": 0.0272,
       "step": 2570
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5229613136654608e-05,
-      "loss": 0.24,
+      "loss": 0.1451,
       "step": 2571
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.522775767696447e-05,
-      "loss": 0.2416,
+      "loss": 0.1005,
       "step": 2572
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.522590221727433e-05,
-      "loss": 0.1387,
+      "loss": 0.0247,
       "step": 2573
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5224046757584192e-05,
-      "loss": 0.289,
+      "loss": 0.0159,
       "step": 2574
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5222191297894056e-05,
-      "loss": 0.0847,
+      "loss": 0.0174,
       "step": 2575
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5220335838203918e-05,
-      "loss": 0.1412,
+      "loss": 0.2246,
       "step": 2576
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5218480378513778e-05,
-      "loss": 0.1937,
+      "loss": 0.1251,
       "step": 2577
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.521662491882364e-05,
-      "loss": 0.1367,
+      "loss": 0.0617,
       "step": 2578
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5214769459133503e-05,
-      "loss": 0.1895,
+      "loss": 0.0596,
       "step": 2579
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5212913999443363e-05,
-      "loss": 0.1395,
+      "loss": 0.1294,
       "step": 2580
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5211058539753225e-05,
-      "loss": 0.1432,
+      "loss": 0.0161,
       "step": 2581
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5209203080063085e-05,
-      "loss": 0.3444,
+      "loss": 0.1527,
       "step": 2582
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5207347620372947e-05,
-      "loss": 0.0889,
+      "loss": 0.0697,
       "step": 2583
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5205492160682811e-05,
-      "loss": 0.1941,
+      "loss": 0.1015,
       "step": 2584
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5203636700992673e-05,
-      "loss": 0.1927,
+      "loss": 0.0131,
       "step": 2585
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5201781241302533e-05,
-      "loss": 0.2423,
+      "loss": 0.0132,
       "step": 2586
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5199925781612395e-05,
-      "loss": 0.0889,
+      "loss": 0.0124,
       "step": 2587
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5198070321922258e-05,
-      "loss": 0.0375,
+      "loss": 0.1623,
       "step": 2588
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5196214862232119e-05,
-      "loss": 0.0375,
+      "loss": 0.0528,
       "step": 2589
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.519435940254198e-05,
-      "loss": 0.1344,
+      "loss": 0.1034,
       "step": 2590
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5192503942851842e-05,
-      "loss": 0.395,
+      "loss": 0.0909,
       "step": 2591
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5190648483161706e-05,
-      "loss": 0.0895,
+      "loss": 0.2021,
       "step": 2592
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5188793023471566e-05,
-      "loss": 0.1387,
+      "loss": 0.0462,
       "step": 2593
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5186937563781428e-05,
-      "loss": 0.1332,
+      "loss": 0.2368,
       "step": 2594
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5185082104091288e-05,
-      "loss": 0.0822,
+      "loss": 0.1104,
       "step": 2595
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5183226644401152e-05,
-      "loss": 0.2414,
+      "loss": 0.1185,
       "step": 2596
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5181371184711014e-05,
-      "loss": 0.1387,
+      "loss": 0.0664,
       "step": 2597
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5179515725020875e-05,
-      "loss": 0.2423,
+      "loss": 0.1054,
       "step": 2598
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5177660265330736e-05,
-      "loss": 0.1943,
+      "loss": 0.0238,
       "step": 2599
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.51758048056406e-05,
-      "loss": 0.1789,
+      "loss": 0.0263,
       "step": 2600
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5173949345950461e-05,
-      "loss": 0.0356,
+      "loss": 0.0835,
       "step": 2601
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5172093886260321e-05,
-      "loss": 0.0375,
+      "loss": 0.0808,
       "step": 2602
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5170238426570183e-05,
-      "loss": 0.088,
+      "loss": 0.0256,
       "step": 2603
     },
     {
       "epoch": 0.72,
       "learning_rate": 1.5168382966880047e-05,
-      "loss": 0.1184,
+      "loss": 0.0641,
       "step": 2604
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5166527507189907e-05,
-      "loss": 0.1381,
+      "loss": 0.1097,
       "step": 2605
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5164672047499769e-05,
-      "loss": 0.086,
+      "loss": 0.0274,
       "step": 2606
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.516281658780963e-05,
-      "loss": 0.0879,
+      "loss": 0.1563,
       "step": 2607
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5160961128119494e-05,
-      "loss": 0.0856,
+      "loss": 0.0642,
       "step": 2608
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5159105668429354e-05,
-      "loss": 0.2099,
+      "loss": 0.1177,
       "step": 2609
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5157250208739216e-05,
-      "loss": 0.082,
+      "loss": 0.244,
       "step": 2610
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5155394749049076e-05,
-      "loss": 0.1378,
+      "loss": 0.2763,
       "step": 2611
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.515353928935894e-05,
-      "loss": 0.1139,
+      "loss": 0.14,
       "step": 2612
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5151683829668802e-05,
-      "loss": 0.1345,
+      "loss": 0.0648,
       "step": 2613
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5149828369978664e-05,
-      "loss": 0.1434,
+      "loss": 0.0847,
       "step": 2614
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5147972910288524e-05,
-      "loss": 0.3562,
+      "loss": 0.0244,
       "step": 2615
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5146117450598387e-05,
-      "loss": 0.0836,
+      "loss": 0.0329,
       "step": 2616
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.514426199090825e-05,
-      "loss": 0.1435,
+      "loss": 0.1227,
       "step": 2617
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.514240653121811e-05,
-      "loss": 0.2011,
+      "loss": 0.0236,
       "step": 2618
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5140551071527971e-05,
-      "loss": 0.0879,
+      "loss": 0.0613,
       "step": 2619
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5138695611837835e-05,
-      "loss": 0.2507,
+      "loss": 0.043,
       "step": 2620
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5136840152147697e-05,
-      "loss": 0.0886,
+      "loss": 0.2793,
       "step": 2621
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5134984692457557e-05,
-      "loss": 0.1404,
+      "loss": 0.1374,
       "step": 2622
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5133129232767419e-05,
-      "loss": 0.1866,
+      "loss": 0.0662,
       "step": 2623
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5131273773077282e-05,
-      "loss": 0.2431,
+      "loss": 0.0599,
       "step": 2624
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5129418313387143e-05,
-      "loss": 0.2771,
+      "loss": 0.0289,
       "step": 2625
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5127562853697004e-05,
-      "loss": 0.1995,
+      "loss": 0.0228,
       "step": 2626
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5125707394006866e-05,
-      "loss": 0.2863,
+      "loss": 0.1013,
       "step": 2627
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5123851934316728e-05,
-      "loss": 0.1091,
+      "loss": 0.0162,
       "step": 2628
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.512199647462659e-05,
-      "loss": 0.0921,
+      "loss": 0.0629,
       "step": 2629
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5120141014936452e-05,
-      "loss": 0.088,
+      "loss": 0.0947,
       "step": 2630
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5118285555246312e-05,
-      "loss": 0.0376,
+      "loss": 0.0533,
       "step": 2631
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5116430095556176e-05,
-      "loss": 0.1372,
+      "loss": 0.0128,
       "step": 2632
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5114574635866037e-05,
-      "loss": 0.2516,
+      "loss": 0.0538,
       "step": 2633
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5112719176175898e-05,
-      "loss": 0.1383,
+      "loss": 0.0954,
       "step": 2634
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.511086371648576e-05,
-      "loss": 0.1907,
+      "loss": 0.0074,
       "step": 2635
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5109008256795623e-05,
-      "loss": 0.1341,
+      "loss": 0.1234,
       "step": 2636
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5107152797105485e-05,
-      "loss": 0.1946,
+      "loss": 0.1768,
       "step": 2637
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5105297337415345e-05,
-      "loss": 0.1485,
+      "loss": 0.0205,
       "step": 2638
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.5103441877725207e-05,
-      "loss": 0.1967,
+      "loss": 0.1876,
       "step": 2639
     },
     {
       "epoch": 0.73,
       "learning_rate": 1.510158641803507e-05,
-      "loss": 0.1925,
+      "loss": 0.0559,
       "step": 2640
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.509973095834493e-05,
-      "loss": 0.1959,
+      "loss": 0.0108,
       "step": 2641
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5097875498654793e-05,
-      "loss": 0.0364,
+      "loss": 0.0858,
       "step": 2642
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5096020038964655e-05,
-      "loss": 0.0862,
+      "loss": 0.0676,
       "step": 2643
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5094164579274518e-05,
-      "loss": 0.1399,
+      "loss": 0.1587,
       "step": 2644
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5092309119584378e-05,
-      "loss": 0.3491,
+      "loss": 0.0647,
       "step": 2645
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.509045365989424e-05,
-      "loss": 0.0887,
+      "loss": 0.0333,
       "step": 2646
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.50885982002041e-05,
-      "loss": 0.1403,
+      "loss": 0.1625,
       "step": 2647
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5086742740513964e-05,
-      "loss": 0.0873,
+      "loss": 0.0215,
       "step": 2648
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5084887280823826e-05,
-      "loss": 0.0876,
+      "loss": 0.1065,
       "step": 2649
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5083031821133688e-05,
-      "loss": 0.1398,
+      "loss": 0.0193,
       "step": 2650
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5081176361443548e-05,
-      "loss": 0.0328,
+      "loss": 0.0743,
       "step": 2651
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5079320901753411e-05,
-      "loss": 0.033,
+      "loss": 0.0143,
       "step": 2652
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5077465442063273e-05,
-      "loss": 0.1944,
+      "loss": 0.0563,
       "step": 2653
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5075609982373133e-05,
-      "loss": 0.1947,
+      "loss": 0.1295,
       "step": 2654
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5073754522682995e-05,
-      "loss": 0.0855,
+      "loss": 0.2078,
       "step": 2655
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5071899062992859e-05,
-      "loss": 0.1412,
+      "loss": 0.0256,
       "step": 2656
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5070043603302719e-05,
-      "loss": 0.3052,
+      "loss": 0.061,
       "step": 2657
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5068188143612581e-05,
-      "loss": 0.1432,
+      "loss": 0.1057,
       "step": 2658
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5066332683922443e-05,
-      "loss": 0.0823,
+      "loss": 0.0269,
       "step": 2659
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5064477224232306e-05,
-      "loss": 0.1386,
+      "loss": 0.0693,
       "step": 2660
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5062621764542166e-05,
-      "loss": 0.0865,
+      "loss": 0.0783,
       "step": 2661
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5060766304852028e-05,
-      "loss": 0.1965,
+      "loss": 0.055,
       "step": 2662
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5058910845161889e-05,
-      "loss": 0.1391,
+      "loss": 0.0659,
       "step": 2663
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5057055385471752e-05,
-      "loss": 0.1409,
+      "loss": 0.0516,
       "step": 2664
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5055199925781614e-05,
-      "loss": 0.141,
+      "loss": 0.0208,
       "step": 2665
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5053344466091476e-05,
-      "loss": 0.3038,
+      "loss": 0.0511,
       "step": 2666
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5051489006401336e-05,
-      "loss": 0.1974,
+      "loss": 0.0569,
       "step": 2667
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.50496335467112e-05,
-      "loss": 0.1931,
+      "loss": 0.1495,
       "step": 2668
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5047778087021061e-05,
-      "loss": 0.0362,
+      "loss": 0.0084,
       "step": 2669
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5045922627330922e-05,
-      "loss": 0.2472,
+      "loss": 0.0592,
       "step": 2670
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5044067167640783e-05,
-      "loss": 0.0359,
+      "loss": 0.1111,
       "step": 2671
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5042211707950647e-05,
-      "loss": 0.1913,
+      "loss": 0.0094,
       "step": 2672
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5040356248260509e-05,
-      "loss": 0.2455,
+      "loss": 0.0751,
       "step": 2673
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5038500788570369e-05,
-      "loss": 0.2499,
+      "loss": 0.0709,
       "step": 2674
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5036645328880231e-05,
-      "loss": 0.0874,
+      "loss": 0.0889,
       "step": 2675
     },
     {
       "epoch": 0.74,
       "learning_rate": 1.5034789869190095e-05,
-      "loss": 0.2952,
+      "loss": 0.1204,
       "step": 2676
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5032934409499955e-05,
-      "loss": 0.2416,
+      "loss": 0.0055,
       "step": 2677
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5031078949809817e-05,
-      "loss": 0.2418,
+      "loss": 0.0044,
       "step": 2678
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5029223490119678e-05,
-      "loss": 0.1892,
+      "loss": 0.137,
       "step": 2679
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.502736803042954e-05,
-      "loss": 0.1393,
+      "loss": 0.1131,
       "step": 2680
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5025512570739402e-05,
-      "loss": 0.288,
+      "loss": 0.0556,
       "step": 2681
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5023657111049264e-05,
-      "loss": 0.2417,
+      "loss": 0.0068,
       "step": 2682
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5021801651359124e-05,
-      "loss": 0.3722,
+      "loss": 0.0932,
       "step": 2683
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5019946191668988e-05,
-      "loss": 0.194,
+      "loss": 0.0588,
       "step": 2684
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.501809073197885e-05,
-      "loss": 0.2367,
+      "loss": 0.0606,
       "step": 2685
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.501623527228871e-05,
-      "loss": 0.2368,
+      "loss": 0.1361,
       "step": 2686
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5014379812598572e-05,
-      "loss": 0.1512,
+      "loss": 0.113,
       "step": 2687
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5012524352908435e-05,
-      "loss": 0.102,
+      "loss": 0.1208,
       "step": 2688
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5010668893218297e-05,
-      "loss": 0.1007,
+      "loss": 0.0911,
       "step": 2689
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5008813433528157e-05,
-      "loss": 0.1888,
+      "loss": 0.0598,
       "step": 2690
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.500695797383802e-05,
-      "loss": 0.1926,
+      "loss": 0.2298,
       "step": 2691
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5005102514147883e-05,
-      "loss": 0.1889,
+      "loss": 0.0928,
       "step": 2692
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5003247054457743e-05,
-      "loss": 0.0954,
+      "loss": 0.0758,
       "step": 2693
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.5001391594767605e-05,
-      "loss": 0.0465,
+      "loss": 0.0585,
       "step": 2694
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4999536135077467e-05,
-      "loss": 0.0929,
+      "loss": 0.1098,
       "step": 2695
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.499768067538733e-05,
-      "loss": 0.0408,
+      "loss": 0.1672,
       "step": 2696
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.499582521569719e-05,
-      "loss": 0.1952,
+      "loss": 0.1087,
       "step": 2697
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4993969756007052e-05,
-      "loss": 0.1413,
+      "loss": 0.1363,
       "step": 2698
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4992114296316912e-05,
-      "loss": 0.1958,
+      "loss": 0.1218,
       "step": 2699
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4990258836626776e-05,
-      "loss": 0.1883,
+      "loss": 0.0994,
       "step": 2700
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4988403376936638e-05,
-      "loss": 0.189,
+      "loss": 0.0695,
       "step": 2701
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4986547917246498e-05,
-      "loss": 0.1458,
+      "loss": 0.102,
       "step": 2702
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.498469245755636e-05,
-      "loss": 0.1394,
+      "loss": 0.0936,
       "step": 2703
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4982836997866224e-05,
-      "loss": 0.0873,
+      "loss": 0.1072,
       "step": 2704
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4980981538176085e-05,
-      "loss": 0.0865,
+      "loss": 0.0587,
       "step": 2705
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4979126078485946e-05,
-      "loss": 0.0329,
+      "loss": 0.0996,
       "step": 2706
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4977270618795807e-05,
-      "loss": 0.0817,
+      "loss": 0.0746,
       "step": 2707
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4975415159105671e-05,
-      "loss": 0.0878,
+      "loss": 0.0204,
       "step": 2708
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4973559699415531e-05,
-      "loss": 0.0305,
+      "loss": 0.1161,
       "step": 2709
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4971704239725393e-05,
-      "loss": 0.1389,
+      "loss": 0.0388,
       "step": 2710
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4969848780035255e-05,
-      "loss": 0.0291,
+      "loss": 0.0554,
       "step": 2711
     },
     {
       "epoch": 0.75,
       "learning_rate": 1.4967993320345118e-05,
-      "loss": 0.1408,
+      "loss": 0.1504,
       "step": 2712
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4966137860654979e-05,
-      "loss": 0.0812,
+      "loss": 0.0414,
       "step": 2713
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.496428240096484e-05,
-      "loss": 0.0822,
+      "loss": 0.0145,
       "step": 2714
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.49624269412747e-05,
-      "loss": 0.0839,
+      "loss": 0.0614,
       "step": 2715
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4960571481584564e-05,
-      "loss": 0.3084,
+      "loss": 0.0384,
       "step": 2716
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4958716021894426e-05,
-      "loss": 0.0251,
+      "loss": 0.0965,
       "step": 2717
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4956860562204288e-05,
-      "loss": 0.0242,
+      "loss": 0.1329,
       "step": 2718
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4955005102514148e-05,
-      "loss": 0.198,
+      "loss": 0.2522,
       "step": 2719
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4953149642824012e-05,
-      "loss": 0.2537,
+      "loss": 0.0123,
       "step": 2720
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4951294183133874e-05,
-      "loss": 0.1943,
+      "loss": 0.102,
       "step": 2721
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4949438723443734e-05,
-      "loss": 0.1361,
+      "loss": 0.0118,
       "step": 2722
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4947583263753596e-05,
-      "loss": 0.1397,
+      "loss": 0.0869,
       "step": 2723
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.494572780406346e-05,
-      "loss": 0.0829,
+      "loss": 0.0844,
       "step": 2724
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.494387234437332e-05,
-      "loss": 0.0247,
+      "loss": 0.1232,
       "step": 2725
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4942016884683181e-05,
-      "loss": 0.1969,
+      "loss": 0.0195,
       "step": 2726
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4940161424993043e-05,
-      "loss": 0.0251,
+      "loss": 0.0268,
       "step": 2727
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4938305965302907e-05,
-      "loss": 0.313,
+      "loss": 0.0603,
       "step": 2728
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4936450505612767e-05,
-      "loss": 0.025,
+      "loss": 0.0231,
       "step": 2729
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4934595045922629e-05,
-      "loss": 0.1407,
+      "loss": 0.055,
       "step": 2730
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4932739586232489e-05,
-      "loss": 0.1393,
+      "loss": 0.1314,
       "step": 2731
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4930884126542353e-05,
-      "loss": 0.0831,
+      "loss": 0.058,
       "step": 2732
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4929028666852214e-05,
-      "loss": 0.1415,
+      "loss": 0.0733,
       "step": 2733
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4927173207162076e-05,
-      "loss": 0.2541,
+      "loss": 0.0127,
       "step": 2734
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4925317747471936e-05,
-      "loss": 0.1413,
+      "loss": 0.1098,
       "step": 2735
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4923462287781798e-05,
-      "loss": 0.1878,
+      "loss": 0.1151,
       "step": 2736
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4921606828091662e-05,
-      "loss": 0.1927,
+      "loss": 0.1191,
       "step": 2737
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4919751368401522e-05,
-      "loss": 0.2496,
+      "loss": 0.094,
       "step": 2738
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4917895908711384e-05,
-      "loss": 0.0836,
+      "loss": 0.1036,
       "step": 2739
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4916040449021246e-05,
-      "loss": 0.2516,
+      "loss": 0.1017,
       "step": 2740
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.491418498933111e-05,
-      "loss": 0.3033,
+      "loss": 0.0981,
       "step": 2741
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.491232952964097e-05,
-      "loss": 0.1335,
+      "loss": 0.162,
       "step": 2742
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4910474069950831e-05,
-      "loss": 0.0855,
+      "loss": 0.1787,
       "step": 2743
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4908618610260692e-05,
-      "loss": 0.1948,
+      "loss": 0.1045,
       "step": 2744
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4906763150570555e-05,
-      "loss": 0.0344,
+      "loss": 0.1231,
       "step": 2745
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4904907690880417e-05,
-      "loss": 0.1904,
+      "loss": 0.0853,
       "step": 2746
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4903052231190279e-05,
-      "loss": 0.0872,
+      "loss": 0.106,
       "step": 2747
     },
     {
       "epoch": 0.76,
       "learning_rate": 1.4901196771500139e-05,
-      "loss": 0.0894,
+      "loss": 0.1516,
       "step": 2748
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4899341311810003e-05,
-      "loss": 0.2414,
+      "loss": 0.1578,
       "step": 2749
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4897485852119864e-05,
-      "loss": 0.1953,
+      "loss": 0.0986,
       "step": 2750
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4895630392429725e-05,
-      "loss": 0.2433,
+      "loss": 0.1234,
       "step": 2751
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4893774932739587e-05,
-      "loss": 0.2423,
+      "loss": 0.0494,
       "step": 2752
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.489191947304945e-05,
-      "loss": 0.1368,
+      "loss": 0.0445,
       "step": 2753
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.489006401335931e-05,
-      "loss": 0.3891,
+      "loss": 0.1085,
       "step": 2754
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4888208553669172e-05,
-      "loss": 0.1875,
+      "loss": 0.067,
       "step": 2755
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4886353093979034e-05,
-      "loss": 0.0921,
+      "loss": 0.0766,
       "step": 2756
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4884497634288898e-05,
-      "loss": 0.1417,
+      "loss": 0.0585,
       "step": 2757
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4882642174598758e-05,
-      "loss": 0.1858,
+      "loss": 0.1575,
       "step": 2758
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.488078671490862e-05,
-      "loss": 0.2429,
+      "loss": 0.0942,
       "step": 2759
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.487893125521848e-05,
-      "loss": 0.2421,
+      "loss": 0.0548,
       "step": 2760
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4877075795528343e-05,
-      "loss": 0.1405,
+      "loss": 0.1302,
       "step": 2761
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4875220335838205e-05,
-      "loss": 0.0456,
+      "loss": 0.1436,
       "step": 2762
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4873364876148067e-05,
-      "loss": 0.1424,
+      "loss": 0.1222,
       "step": 2763
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4871509416457927e-05,
-      "loss": 0.1463,
+      "loss": 0.1879,
       "step": 2764
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4869653956767791e-05,
-      "loss": 0.2332,
+      "loss": 0.0182,
       "step": 2765
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4867798497077653e-05,
-      "loss": 0.0939,
+      "loss": 0.0532,
       "step": 2766
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4865943037387513e-05,
-      "loss": 0.0941,
+      "loss": 0.0697,
       "step": 2767
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4864087577697375e-05,
-      "loss": 0.1456,
+      "loss": 0.1437,
       "step": 2768
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4862232118007238e-05,
-      "loss": 0.2421,
+      "loss": 0.062,
       "step": 2769
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.48603766583171e-05,
-      "loss": 0.1796,
+      "loss": 0.177,
       "step": 2770
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.485852119862696e-05,
-      "loss": 0.194,
+      "loss": 0.1267,
       "step": 2771
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4856665738936822e-05,
-      "loss": 0.093,
+      "loss": 0.0235,
       "step": 2772
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4854810279246686e-05,
-      "loss": 0.1901,
+      "loss": 0.0641,
       "step": 2773
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4852954819556546e-05,
-      "loss": 0.2387,
+      "loss": 0.0222,
       "step": 2774
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4851099359866408e-05,
-      "loss": 0.1901,
+      "loss": 0.2195,
       "step": 2775
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.484924390017627e-05,
-      "loss": 0.0917,
+      "loss": 0.0254,
       "step": 2776
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4847388440486132e-05,
-      "loss": 0.0902,
+      "loss": 0.0244,
       "step": 2777
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4845532980795993e-05,
-      "loss": 0.1429,
+      "loss": 0.0821,
       "step": 2778
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4843677521105855e-05,
-      "loss": 0.1838,
+      "loss": 0.1392,
       "step": 2779
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4841822061415716e-05,
-      "loss": 0.2428,
+      "loss": 0.0257,
       "step": 2780
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4839966601725579e-05,
-      "loss": 0.0909,
+      "loss": 0.1406,
       "step": 2781
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4838111142035441e-05,
-      "loss": 0.1916,
+      "loss": 0.2371,
       "step": 2782
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4836255682345301e-05,
-      "loss": 0.1379,
+      "loss": 0.0833,
       "step": 2783
     },
     {
       "epoch": 0.77,
       "learning_rate": 1.4834400222655163e-05,
-      "loss": 0.1382,
+      "loss": 0.0772,
       "step": 2784
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4832544762965027e-05,
-      "loss": 0.1379,
+      "loss": 0.0521,
       "step": 2785
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4830689303274888e-05,
-      "loss": 0.1382,
+      "loss": 0.1666,
       "step": 2786
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4828833843584749e-05,
-      "loss": 0.249,
+      "loss": 0.066,
       "step": 2787
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.482697838389461e-05,
-      "loss": 0.2467,
+      "loss": 0.1656,
       "step": 2788
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4825122924204474e-05,
-      "loss": 0.1405,
+      "loss": 0.0215,
       "step": 2789
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4823267464514334e-05,
-      "loss": 0.0849,
+      "loss": 0.1187,
       "step": 2790
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4821412004824196e-05,
-      "loss": 0.189,
+      "loss": 0.0971,
       "step": 2791
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4819556545134058e-05,
-      "loss": 0.1369,
+      "loss": 0.1006,
       "step": 2792
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4817701085443922e-05,
-      "loss": 0.0903,
+      "loss": 0.0709,
       "step": 2793
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4815845625753782e-05,
-      "loss": 0.1461,
+      "loss": 0.0732,
       "step": 2794
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4813990166063644e-05,
-      "loss": 0.1425,
+      "loss": 0.0302,
       "step": 2795
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4812134706373504e-05,
-      "loss": 0.1897,
+      "loss": 0.1225,
       "step": 2796
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4810279246683367e-05,
-      "loss": 0.2481,
+      "loss": 0.1012,
       "step": 2797
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.480842378699323e-05,
-      "loss": 0.0888,
+      "loss": 0.0717,
       "step": 2798
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4806568327303091e-05,
-      "loss": 0.1908,
+      "loss": 0.0242,
       "step": 2799
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4804712867612951e-05,
-      "loss": 0.1396,
+      "loss": 0.0879,
       "step": 2800
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4802857407922815e-05,
-      "loss": 0.089,
+      "loss": 0.0246,
       "step": 2801
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4801001948232677e-05,
-      "loss": 0.1426,
+      "loss": 0.0207,
       "step": 2802
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4799146488542537e-05,
-      "loss": 0.0376,
+      "loss": 0.0981,
       "step": 2803
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4797291028852399e-05,
-      "loss": 0.1947,
+      "loss": 0.0475,
       "step": 2804
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4795435569162262e-05,
-      "loss": 0.2463,
+      "loss": 0.0158,
       "step": 2805
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4793580109472122e-05,
-      "loss": 0.1407,
+      "loss": 0.0166,
       "step": 2806
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4791724649781984e-05,
-      "loss": 0.0869,
+      "loss": 0.0821,
       "step": 2807
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4789869190091846e-05,
-      "loss": 0.2908,
+      "loss": 0.0102,
       "step": 2808
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.478801373040171e-05,
-      "loss": 0.0892,
+      "loss": 0.1325,
       "step": 2809
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.478615827071157e-05,
-      "loss": 0.0864,
+      "loss": 0.1455,
       "step": 2810
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4784302811021432e-05,
-      "loss": 0.1403,
+      "loss": 0.2387,
       "step": 2811
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4782447351331292e-05,
-      "loss": 0.1372,
+      "loss": 0.0079,
       "step": 2812
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4780591891641156e-05,
-      "loss": 0.0349,
+      "loss": 0.0618,
       "step": 2813
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4778736431951017e-05,
-      "loss": 0.2943,
+      "loss": 0.0721,
       "step": 2814
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.477688097226088e-05,
-      "loss": 0.2447,
+      "loss": 0.1241,
       "step": 2815
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.477502551257074e-05,
-      "loss": 0.0867,
+      "loss": 0.097,
       "step": 2816
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4773170052880603e-05,
-      "loss": 0.0883,
+      "loss": 0.1128,
       "step": 2817
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4771314593190465e-05,
-      "loss": 0.0874,
+      "loss": 0.1021,
       "step": 2818
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4769459133500325e-05,
-      "loss": 0.0333,
+      "loss": 0.0527,
       "step": 2819
     },
     {
       "epoch": 0.78,
       "learning_rate": 1.4767603673810187e-05,
-      "loss": 0.1413,
+      "loss": 0.053,
       "step": 2820
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.476574821412005e-05,
-      "loss": 0.2465,
+      "loss": 0.1306,
       "step": 2821
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.476389275442991e-05,
-      "loss": 0.0816,
+      "loss": 0.1393,
       "step": 2822
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4762037294739773e-05,
-      "loss": 0.2477,
+      "loss": 0.0884,
       "step": 2823
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4760181835049634e-05,
-      "loss": 0.1365,
+      "loss": 0.1444,
       "step": 2824
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4758326375359498e-05,
-      "loss": 0.1857,
+      "loss": 0.0995,
       "step": 2825
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4756470915669358e-05,
-      "loss": 0.1393,
+      "loss": 0.0388,
       "step": 2826
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.475461545597922e-05,
-      "loss": 0.1992,
+      "loss": 0.1283,
       "step": 2827
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.475275999628908e-05,
-      "loss": 0.1944,
+      "loss": 0.1159,
       "step": 2828
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4750904536598944e-05,
-      "loss": 0.1906,
+      "loss": 0.0712,
       "step": 2829
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4749049076908806e-05,
-      "loss": 0.1377,
+      "loss": 0.0998,
       "step": 2830
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4747193617218668e-05,
-      "loss": 0.3009,
+      "loss": 0.0329,
       "step": 2831
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4745338157528528e-05,
-      "loss": 0.1947,
+      "loss": 0.0319,
       "step": 2832
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4743482697838391e-05,
-      "loss": 0.1391,
+      "loss": 0.2335,
       "step": 2833
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4741627238148253e-05,
-      "loss": 0.2434,
+      "loss": 0.0933,
       "step": 2834
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4739771778458113e-05,
-      "loss": 0.0357,
+      "loss": 0.09,
       "step": 2835
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4737916318767975e-05,
-      "loss": 0.2902,
+      "loss": 0.0693,
       "step": 2836
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4736060859077839e-05,
-      "loss": 0.087,
+      "loss": 0.063,
       "step": 2837
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.47342053993877e-05,
-      "loss": 0.0871,
+      "loss": 0.0166,
       "step": 2838
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.473234993969756e-05,
-      "loss": 0.0912,
+      "loss": 0.0823,
       "step": 2839
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4730494480007423e-05,
-      "loss": 0.1394,
+      "loss": 0.1181,
       "step": 2840
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4728639020317286e-05,
-      "loss": 0.2413,
+      "loss": 0.0491,
       "step": 2841
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4726783560627146e-05,
-      "loss": 0.2916,
+      "loss": 0.0193,
       "step": 2842
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4724928100937008e-05,
-      "loss": 0.2425,
+      "loss": 0.2266,
       "step": 2843
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.472307264124687e-05,
-      "loss": 0.0881,
+      "loss": 0.0851,
       "step": 2844
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4721217181556732e-05,
-      "loss": 0.1407,
+      "loss": 0.0686,
       "step": 2845
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4719361721866594e-05,
-      "loss": 0.194,
+      "loss": 0.1038,
       "step": 2846
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4717506262176456e-05,
-      "loss": 0.1974,
+      "loss": 0.0272,
       "step": 2847
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4715650802486316e-05,
-      "loss": 0.1435,
+      "loss": 0.0776,
       "step": 2848
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.471379534279618e-05,
-      "loss": 0.1925,
+      "loss": 0.0564,
       "step": 2849
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4711939883106041e-05,
-      "loss": 0.0418,
+      "loss": 0.0683,
       "step": 2850
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4710084423415902e-05,
-      "loss": 0.1426,
+      "loss": 0.1127,
       "step": 2851
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4708228963725763e-05,
-      "loss": 0.0406,
+      "loss": 0.1237,
       "step": 2852
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4706373504035627e-05,
-      "loss": 0.0915,
+      "loss": 0.0326,
       "step": 2853
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4704518044345489e-05,
-      "loss": 0.0912,
+      "loss": 0.0487,
       "step": 2854
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4702662584655349e-05,
-      "loss": 0.3952,
+      "loss": 0.0331,
       "step": 2855
     },
     {
       "epoch": 0.79,
       "learning_rate": 1.4700807124965211e-05,
-      "loss": 0.039,
+      "loss": 0.1299,
       "step": 2856
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4698951665275074e-05,
-      "loss": 0.1922,
+      "loss": 0.2108,
       "step": 2857
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4697096205584935e-05,
-      "loss": 0.089,
+      "loss": 0.0158,
       "step": 2858
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4695240745894797e-05,
-      "loss": 0.3372,
+      "loss": 0.2149,
       "step": 2859
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4693385286204658e-05,
-      "loss": 0.1413,
+      "loss": 0.071,
       "step": 2860
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4691529826514522e-05,
-      "loss": 0.1881,
+      "loss": 0.1018,
       "step": 2861
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4689674366824382e-05,
-      "loss": 0.353,
+      "loss": 0.0637,
       "step": 2862
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4687818907134244e-05,
-      "loss": 0.3994,
+      "loss": 0.0239,
       "step": 2863
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4685963447444104e-05,
-      "loss": 0.1395,
+      "loss": 0.068,
       "step": 2864
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4684107987753968e-05,
-      "loss": 0.139,
+      "loss": 0.0625,
       "step": 2865
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.468225252806383e-05,
-      "loss": 0.1388,
+      "loss": 0.059,
       "step": 2866
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4680397068373691e-05,
-      "loss": 0.1395,
+      "loss": 0.1227,
       "step": 2867
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4678541608683552e-05,
-      "loss": 0.0923,
+      "loss": 0.0791,
       "step": 2868
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4676686148993415e-05,
-      "loss": 0.0896,
+      "loss": 0.0603,
       "step": 2869
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4674830689303277e-05,
-      "loss": 0.1882,
+      "loss": 0.1224,
       "step": 2870
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4672975229613137e-05,
-      "loss": 0.0926,
+      "loss": 0.0142,
       "step": 2871
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4671119769922999e-05,
-      "loss": 0.2896,
+      "loss": 0.1475,
       "step": 2872
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4669264310232863e-05,
-      "loss": 0.1902,
+      "loss": 0.0658,
       "step": 2873
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4667408850542723e-05,
-      "loss": 0.1837,
+      "loss": 0.0847,
       "step": 2874
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4665553390852585e-05,
-      "loss": 0.0908,
+      "loss": 0.0158,
       "step": 2875
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4663697931162447e-05,
-      "loss": 0.2404,
+      "loss": 0.0653,
       "step": 2876
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.466184247147231e-05,
-      "loss": 0.2369,
+      "loss": 0.1096,
       "step": 2877
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.465998701178217e-05,
-      "loss": 0.2337,
+      "loss": 0.1161,
       "step": 2878
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4658131552092032e-05,
-      "loss": 0.1873,
+      "loss": 0.0874,
       "step": 2879
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4656276092401892e-05,
-      "loss": 0.091,
+      "loss": 0.1797,
       "step": 2880
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4654420632711756e-05,
-      "loss": 0.2354,
+      "loss": 0.1633,
       "step": 2881
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4652565173021618e-05,
-      "loss": 0.1346,
+      "loss": 0.2145,
       "step": 2882
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.465070971333148e-05,
-      "loss": 0.1388,
+      "loss": 0.0694,
       "step": 2883
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.464885425364134e-05,
-      "loss": 0.1919,
+      "loss": 0.1041,
       "step": 2884
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4646998793951202e-05,
-      "loss": 0.0915,
+      "loss": 0.0715,
       "step": 2885
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4645143334261065e-05,
-      "loss": 0.1398,
+      "loss": 0.0396,
       "step": 2886
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4643287874570926e-05,
-      "loss": 0.1405,
+      "loss": 0.0615,
       "step": 2887
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4641432414880787e-05,
-      "loss": 0.0891,
+      "loss": 0.2253,
       "step": 2888
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.463957695519065e-05,
-      "loss": 0.0914,
+      "loss": 0.0781,
       "step": 2889
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4637721495500513e-05,
-      "loss": 0.1384,
+      "loss": 0.0679,
       "step": 2890
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4635866035810373e-05,
-      "loss": 0.141,
+      "loss": 0.1249,
       "step": 2891
     },
     {
       "epoch": 0.8,
       "learning_rate": 1.4634010576120235e-05,
-      "loss": 0.1401,
+      "loss": 0.1139,
       "step": 2892
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4632155116430095e-05,
-      "loss": 0.1916,
+      "loss": 0.0699,
       "step": 2893
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4630299656739959e-05,
-      "loss": 0.1933,
+      "loss": 0.0883,
       "step": 2894
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.462844419704982e-05,
-      "loss": 0.1417,
+      "loss": 0.0242,
       "step": 2895
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4626588737359682e-05,
-      "loss": 0.1356,
+      "loss": 0.0358,
       "step": 2896
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4624733277669543e-05,
-      "loss": 0.0869,
+      "loss": 0.1134,
       "step": 2897
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4622877817979406e-05,
-      "loss": 0.1396,
+      "loss": 0.1039,
       "step": 2898
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4621022358289268e-05,
-      "loss": 0.1402,
+      "loss": 0.0558,
       "step": 2899
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4619166898599128e-05,
-      "loss": 0.1899,
+      "loss": 0.0609,
       "step": 2900
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.461731143890899e-05,
-      "loss": 0.1953,
+      "loss": 0.0153,
       "step": 2901
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4615455979218854e-05,
-      "loss": 0.0901,
+      "loss": 0.0144,
       "step": 2902
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4613600519528714e-05,
-      "loss": 0.3465,
+      "loss": 0.0536,
       "step": 2903
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4611745059838576e-05,
-      "loss": 0.1927,
+      "loss": 0.1043,
       "step": 2904
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4609889600148437e-05,
-      "loss": 0.0865,
+      "loss": 0.0143,
       "step": 2905
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4608034140458301e-05,
-      "loss": 0.091,
+      "loss": 0.0434,
       "step": 2906
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4606178680768161e-05,
-      "loss": 0.1457,
+      "loss": 0.0268,
       "step": 2907
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4604323221078023e-05,
-      "loss": 0.1981,
+      "loss": 0.0062,
       "step": 2908
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4602467761387883e-05,
-      "loss": 0.1457,
+      "loss": 0.0059,
       "step": 2909
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4600612301697747e-05,
-      "loss": 0.1455,
+      "loss": 0.0072,
       "step": 2910
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4598756842007609e-05,
-      "loss": 0.1928,
+      "loss": 0.0277,
       "step": 2911
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.459690138231747e-05,
-      "loss": 0.2857,
+      "loss": 0.0031,
       "step": 2912
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.459504592262733e-05,
-      "loss": 0.1912,
+      "loss": 0.0513,
       "step": 2913
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4593190462937194e-05,
-      "loss": 0.0354,
+      "loss": 0.1785,
       "step": 2914
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4591335003247056e-05,
-      "loss": 0.2913,
+      "loss": 0.1587,
       "step": 2915
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4589479543556916e-05,
-      "loss": 0.0362,
+      "loss": 0.0038,
       "step": 2916
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4587624083866778e-05,
-      "loss": 0.0896,
+      "loss": 0.0027,
       "step": 2917
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4585768624176642e-05,
-      "loss": 0.198,
+      "loss": 0.053,
       "step": 2918
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4583913164486504e-05,
-      "loss": 0.1901,
+      "loss": 0.0285,
       "step": 2919
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4582057704796364e-05,
-      "loss": 0.0362,
+      "loss": 0.1312,
       "step": 2920
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4580202245106226e-05,
-      "loss": 0.0875,
+      "loss": 0.0502,
       "step": 2921
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.457834678541609e-05,
-      "loss": 0.0858,
+      "loss": 0.0996,
       "step": 2922
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.457649132572595e-05,
-      "loss": 0.1393,
+      "loss": 0.1783,
       "step": 2923
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4574635866035811e-05,
-      "loss": 0.0841,
+      "loss": 0.1451,
       "step": 2924
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4572780406345673e-05,
-      "loss": 0.0879,
+      "loss": 0.1479,
       "step": 2925
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4570924946655535e-05,
-      "loss": 0.0863,
+      "loss": 0.1067,
       "step": 2926
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4569069486965397e-05,
-      "loss": 0.1421,
+      "loss": 0.1054,
       "step": 2927
     },
     {
       "epoch": 0.81,
       "learning_rate": 1.4567214027275259e-05,
-      "loss": 0.0868,
+      "loss": 0.0665,
       "step": 2928
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4565358567585119e-05,
-      "loss": 0.0855,
+      "loss": 0.0356,
       "step": 2929
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4563503107894983e-05,
-      "loss": 0.0845,
+      "loss": 0.097,
       "step": 2930
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4561647648204844e-05,
-      "loss": 0.032,
+      "loss": 0.1327,
       "step": 2931
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4559792188514705e-05,
-      "loss": 0.2495,
+      "loss": 0.0635,
       "step": 2932
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4557936728824566e-05,
-      "loss": 0.1973,
+      "loss": 0.0358,
       "step": 2933
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.455608126913443e-05,
-      "loss": 0.0848,
+      "loss": 0.0765,
       "step": 2934
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4554225809444292e-05,
-      "loss": 0.1379,
+      "loss": 0.038,
       "step": 2935
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4552370349754152e-05,
-      "loss": 0.1983,
+      "loss": 0.177,
       "step": 2936
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4550514890064014e-05,
-      "loss": 0.1433,
+      "loss": 0.0226,
       "step": 2937
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4548659430373878e-05,
-      "loss": 0.2501,
+      "loss": 0.0184,
       "step": 2938
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4546803970683738e-05,
-      "loss": 0.027,
+      "loss": 0.0865,
       "step": 2939
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.45449485109936e-05,
-      "loss": 0.135,
+      "loss": 0.0691,
       "step": 2940
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4543093051303461e-05,
-      "loss": 0.0849,
+      "loss": 0.02,
       "step": 2941
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4541237591613323e-05,
-      "loss": 0.029,
+      "loss": 0.2262,
       "step": 2942
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4539382131923185e-05,
-      "loss": 0.1934,
+      "loss": 0.1263,
       "step": 2943
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4537526672233047e-05,
-      "loss": 0.1962,
+      "loss": 0.0298,
       "step": 2944
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4535671212542907e-05,
-      "loss": 0.0806,
+      "loss": 0.0182,
       "step": 2945
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.453381575285277e-05,
-      "loss": 0.1371,
+      "loss": 0.0607,
       "step": 2946
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4531960293162633e-05,
-      "loss": 0.0275,
+      "loss": 0.0684,
       "step": 2947
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4530104833472493e-05,
-      "loss": 0.138,
+      "loss": 0.0375,
       "step": 2948
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4528249373782355e-05,
-      "loss": 0.0846,
+      "loss": 0.1294,
       "step": 2949
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4526393914092218e-05,
-      "loss": 0.3042,
+      "loss": 0.0149,
       "step": 2950
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.452453845440208e-05,
-      "loss": 0.1924,
+      "loss": 0.0636,
       "step": 2951
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.452268299471194e-05,
-      "loss": 0.0836,
+      "loss": 0.1405,
       "step": 2952
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4520827535021802e-05,
-      "loss": 0.1976,
+      "loss": 0.1297,
       "step": 2953
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4518972075331666e-05,
-      "loss": 0.1355,
+      "loss": 0.1465,
       "step": 2954
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4517116615641526e-05,
-      "loss": 0.311,
+      "loss": 0.0638,
       "step": 2955
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4515261155951388e-05,
-      "loss": 0.1437,
+      "loss": 0.0141,
       "step": 2956
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.451340569626125e-05,
-      "loss": 0.0838,
+      "loss": 0.0943,
       "step": 2957
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4511550236571113e-05,
-      "loss": 0.2563,
+      "loss": 0.0565,
       "step": 2958
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4509694776880973e-05,
-      "loss": 0.0862,
+      "loss": 0.0143,
       "step": 2959
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4507839317190835e-05,
-      "loss": 0.0875,
+      "loss": 0.1306,
       "step": 2960
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4505983857500695e-05,
-      "loss": 0.1326,
+      "loss": 0.0216,
       "step": 2961
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4504128397810559e-05,
-      "loss": 0.0863,
+      "loss": 0.0479,
       "step": 2962
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4502272938120421e-05,
-      "loss": 0.1932,
+      "loss": 0.0608,
       "step": 2963
     },
     {
       "epoch": 0.82,
       "learning_rate": 1.4500417478430283e-05,
-      "loss": 0.2447,
+      "loss": 0.0794,
       "step": 2964
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4498562018740143e-05,
-      "loss": 0.1969,
+      "loss": 0.0197,
       "step": 2965
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4496706559050007e-05,
-      "loss": 0.2432,
+      "loss": 0.0802,
       "step": 2966
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4494851099359868e-05,
-      "loss": 0.2453,
+      "loss": 0.0638,
       "step": 2967
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4492995639669729e-05,
-      "loss": 0.1894,
+      "loss": 0.0419,
       "step": 2968
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.449114017997959e-05,
-      "loss": 0.2929,
+      "loss": 0.1068,
       "step": 2969
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4489284720289454e-05,
-      "loss": 0.1389,
+      "loss": 0.0341,
       "step": 2970
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4487429260599314e-05,
-      "loss": 0.1407,
+      "loss": 0.0139,
       "step": 2971
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4485573800909176e-05,
-      "loss": 0.0383,
+      "loss": 0.0654,
       "step": 2972
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4483718341219038e-05,
-      "loss": 0.1424,
+      "loss": 0.0165,
       "step": 2973
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4481862881528901e-05,
-      "loss": 0.0905,
+      "loss": 0.0142,
       "step": 2974
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4480007421838762e-05,
-      "loss": 0.0915,
+      "loss": 0.0926,
       "step": 2975
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4478151962148624e-05,
-      "loss": 0.1872,
+      "loss": 0.163,
       "step": 2976
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4476296502458484e-05,
-      "loss": 0.1901,
+      "loss": 0.2603,
       "step": 2977
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4474441042768347e-05,
-      "loss": 0.0399,
+      "loss": 0.0651,
       "step": 2978
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4472585583078209e-05,
-      "loss": 0.0886,
+      "loss": 0.0152,
       "step": 2979
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4470730123388071e-05,
-      "loss": 0.0882,
+      "loss": 0.0095,
       "step": 2980
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4468874663697931e-05,
-      "loss": 0.2362,
+      "loss": 0.0622,
       "step": 2981
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4467019204007795e-05,
-      "loss": 0.0881,
+      "loss": 0.0176,
       "step": 2982
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4465163744317657e-05,
-      "loss": 0.1905,
+      "loss": 0.2451,
       "step": 2983
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4463308284627517e-05,
-      "loss": 0.1382,
+      "loss": 0.2832,
       "step": 2984
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4461452824937379e-05,
-      "loss": 0.1872,
+      "loss": 0.0596,
       "step": 2985
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4459597365247242e-05,
-      "loss": 0.0856,
+      "loss": 0.1179,
       "step": 2986
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4457741905557104e-05,
-      "loss": 0.246,
+      "loss": 0.0575,
       "step": 2987
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4455886445866964e-05,
-      "loss": 0.2428,
+      "loss": 0.0221,
       "step": 2988
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4454030986176826e-05,
-      "loss": 0.1377,
+      "loss": 0.0947,
       "step": 2989
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.445217552648669e-05,
-      "loss": 0.1916,
+      "loss": 0.1205,
       "step": 2990
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.445032006679655e-05,
-      "loss": 0.0903,
+      "loss": 0.1733,
       "step": 2991
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4448464607106412e-05,
-      "loss": 0.2448,
+      "loss": 0.1546,
       "step": 2992
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4446609147416274e-05,
-      "loss": 0.0368,
+      "loss": 0.0654,
       "step": 2993
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4444753687726136e-05,
-      "loss": 0.241,
+      "loss": 0.0202,
       "step": 2994
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4442898228035997e-05,
-      "loss": 0.1376,
+      "loss": 0.1414,
       "step": 2995
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.444104276834586e-05,
-      "loss": 0.2428,
+      "loss": 0.0694,
       "step": 2996
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.443918730865572e-05,
-      "loss": 0.1936,
+      "loss": 0.0684,
       "step": 2997
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4437331848965583e-05,
-      "loss": 0.3398,
+      "loss": 0.0526,
       "step": 2998
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4435476389275445e-05,
-      "loss": 0.092,
+      "loss": 0.0677,
       "step": 2999
     },
     {
       "epoch": 0.83,
       "learning_rate": 1.4433620929585305e-05,
-      "loss": 0.2434,
+      "loss": 0.0257,
       "step": 3000
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4431765469895167e-05,
-      "loss": 0.2919,
+      "loss": 0.2032,
       "step": 3001
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.442991001020503e-05,
-      "loss": 0.1417,
+      "loss": 0.073,
       "step": 3002
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4428054550514892e-05,
-      "loss": 0.0943,
+      "loss": 0.0699,
       "step": 3003
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4426199090824753e-05,
-      "loss": 0.0946,
+      "loss": 0.0326,
       "step": 3004
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4424343631134614e-05,
-      "loss": 0.1385,
+      "loss": 0.0971,
       "step": 3005
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4422488171444478e-05,
-      "loss": 0.186,
+      "loss": 0.08,
       "step": 3006
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4420632711754338e-05,
-      "loss": 0.3847,
+      "loss": 0.0822,
       "step": 3007
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.44187772520642e-05,
-      "loss": 0.2404,
+      "loss": 0.0194,
       "step": 3008
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4416921792374062e-05,
-      "loss": 0.192,
+      "loss": 0.0202,
       "step": 3009
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4415066332683925e-05,
-      "loss": 0.2309,
+      "loss": 0.0139,
       "step": 3010
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4413210872993786e-05,
-      "loss": 0.1395,
+      "loss": 0.0753,
       "step": 3011
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4411355413303647e-05,
-      "loss": 0.326,
+      "loss": 0.0593,
       "step": 3012
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4409499953613508e-05,
-      "loss": 0.0541,
+      "loss": 0.1633,
       "step": 3013
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4407644493923371e-05,
-      "loss": 0.1001,
+      "loss": 0.0137,
       "step": 3014
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4405789034233233e-05,
-      "loss": 0.0944,
+      "loss": 0.0916,
       "step": 3015
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4403933574543095e-05,
-      "loss": 0.1889,
+      "loss": 0.0691,
       "step": 3016
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4402078114852955e-05,
-      "loss": 0.1934,
+      "loss": 0.1383,
       "step": 3017
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4400222655162819e-05,
-      "loss": 0.0976,
+      "loss": 0.0116,
       "step": 3018
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.439836719547268e-05,
-      "loss": 0.1405,
+      "loss": 0.0823,
       "step": 3019
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.439651173578254e-05,
-      "loss": 0.2891,
+      "loss": 0.0166,
       "step": 3020
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4394656276092403e-05,
-      "loss": 0.1898,
+      "loss": 0.0203,
       "step": 3021
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4392800816402266e-05,
-      "loss": 0.0939,
+      "loss": 0.066,
       "step": 3022
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4390945356712126e-05,
-      "loss": 0.2785,
+      "loss": 0.0688,
       "step": 3023
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4389089897021988e-05,
-      "loss": 0.1375,
+      "loss": 0.045,
       "step": 3024
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.438723443733185e-05,
-      "loss": 0.0906,
+      "loss": 0.0139,
       "step": 3025
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4385378977641714e-05,
-      "loss": 0.1425,
+      "loss": 0.2548,
       "step": 3026
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4383523517951574e-05,
-      "loss": 0.0897,
+      "loss": 0.1146,
       "step": 3027
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4381668058261436e-05,
-      "loss": 0.1868,
+      "loss": 0.0682,
       "step": 3028
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4379812598571296e-05,
-      "loss": 0.1368,
+      "loss": 0.1651,
       "step": 3029
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.437795713888116e-05,
-      "loss": 0.0883,
+      "loss": 0.0134,
       "step": 3030
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4376101679191021e-05,
-      "loss": 0.1417,
+      "loss": 0.062,
       "step": 3031
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4374246219500883e-05,
-      "loss": 0.2384,
+      "loss": 0.1078,
       "step": 3032
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4372390759810743e-05,
-      "loss": 0.0905,
+      "loss": 0.0146,
       "step": 3033
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4370535300120607e-05,
-      "loss": 0.3474,
+      "loss": 0.0592,
       "step": 3034
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4368679840430469e-05,
-      "loss": 0.0884,
+      "loss": 0.2167,
       "step": 3035
     },
     {
       "epoch": 0.84,
       "learning_rate": 1.4366824380740329e-05,
-      "loss": 0.0868,
+      "loss": 0.1249,
       "step": 3036
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4364968921050191e-05,
-      "loss": 0.1894,
+      "loss": 0.1364,
       "step": 3037
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4363113461360053e-05,
-      "loss": 0.0869,
+      "loss": 0.1656,
       "step": 3038
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4361258001669916e-05,
-      "loss": 0.0371,
+      "loss": 0.0212,
       "step": 3039
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4359402541979776e-05,
-      "loss": 0.3974,
+      "loss": 0.1067,
       "step": 3040
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4357547082289638e-05,
-      "loss": 0.0888,
+      "loss": 0.1033,
       "step": 3041
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4355691622599499e-05,
-      "loss": 0.0868,
+      "loss": 0.0202,
       "step": 3042
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4353836162909362e-05,
-      "loss": 0.0891,
+      "loss": 0.0679,
       "step": 3043
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4351980703219224e-05,
-      "loss": 0.1974,
+      "loss": 0.1174,
       "step": 3044
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4350125243529086e-05,
-      "loss": 0.1921,
+      "loss": 0.0728,
       "step": 3045
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4348269783838946e-05,
-      "loss": 0.2398,
+      "loss": 0.0295,
       "step": 3046
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.434641432414881e-05,
-      "loss": 0.1899,
+      "loss": 0.0798,
       "step": 3047
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4344558864458671e-05,
-      "loss": 0.0343,
+      "loss": 0.0308,
       "step": 3048
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4342703404768532e-05,
-      "loss": 0.1956,
+      "loss": 0.1334,
       "step": 3049
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4340847945078393e-05,
-      "loss": 0.0875,
+      "loss": 0.0573,
       "step": 3050
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4338992485388257e-05,
-      "loss": 0.1938,
+      "loss": 0.1212,
       "step": 3051
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4337137025698117e-05,
-      "loss": 0.0846,
+      "loss": 0.0346,
       "step": 3052
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4335281566007979e-05,
-      "loss": 0.0871,
+      "loss": 0.1231,
       "step": 3053
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4333426106317841e-05,
-      "loss": 0.3484,
+      "loss": 0.0372,
       "step": 3054
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4331570646627705e-05,
-      "loss": 0.1455,
+      "loss": 0.0101,
       "step": 3055
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4329715186937565e-05,
-      "loss": 0.135,
+      "loss": 0.0666,
       "step": 3056
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4327859727247427e-05,
-      "loss": 0.1948,
+      "loss": 0.0132,
       "step": 3057
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4326004267557287e-05,
-      "loss": 0.4538,
+      "loss": 0.1643,
       "step": 3058
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.432414880786715e-05,
-      "loss": 0.2957,
+      "loss": 0.0145,
       "step": 3059
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4322293348177012e-05,
-      "loss": 0.0355,
+      "loss": 0.2079,
       "step": 3060
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4320437888486874e-05,
-      "loss": 0.0893,
+      "loss": 0.0117,
       "step": 3061
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4318582428796734e-05,
-      "loss": 0.0372,
+      "loss": 0.0612,
       "step": 3062
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4316726969106598e-05,
-      "loss": 0.242,
+      "loss": 0.1083,
       "step": 3063
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.431487150941646e-05,
-      "loss": 0.1916,
+      "loss": 0.0232,
       "step": 3064
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.431301604972632e-05,
-      "loss": 0.0903,
+      "loss": 0.1144,
       "step": 3065
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4311160590036182e-05,
-      "loss": 0.1411,
+      "loss": 0.0543,
       "step": 3066
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4309305130346045e-05,
-      "loss": 0.0909,
+      "loss": 0.0719,
       "step": 3067
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4307449670655905e-05,
-      "loss": 0.0871,
+      "loss": 0.0088,
       "step": 3068
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4305594210965767e-05,
-      "loss": 0.1884,
+      "loss": 0.0091,
       "step": 3069
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.430373875127563e-05,
-      "loss": 0.2476,
+      "loss": 0.0089,
       "step": 3070
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4301883291585493e-05,
-      "loss": 0.1894,
+      "loss": 0.1489,
       "step": 3071
     },
     {
       "epoch": 0.85,
       "learning_rate": 1.4300027831895353e-05,
-      "loss": 0.1907,
+      "loss": 0.2164,
       "step": 3072
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4298172372205215e-05,
-      "loss": 0.089,
+      "loss": 0.2021,
       "step": 3073
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4296316912515075e-05,
-      "loss": 0.287,
+      "loss": 0.1237,
       "step": 3074
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4294461452824939e-05,
-      "loss": 0.1893,
+      "loss": 0.0583,
       "step": 3075
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.42926059931348e-05,
-      "loss": 0.1909,
+      "loss": 0.0953,
       "step": 3076
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4290750533444662e-05,
-      "loss": 0.1408,
+      "loss": 0.2002,
       "step": 3077
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4288895073754522e-05,
-      "loss": 0.1891,
+      "loss": 0.0122,
       "step": 3078
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4287039614064386e-05,
-      "loss": 0.2385,
+      "loss": 0.0692,
       "step": 3079
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4285184154374248e-05,
-      "loss": 0.1398,
+      "loss": 0.0161,
       "step": 3080
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4283328694684108e-05,
-      "loss": 0.2386,
+      "loss": 0.1243,
       "step": 3081
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.428147323499397e-05,
-      "loss": 0.294,
+      "loss": 0.0936,
       "step": 3082
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4279617775303834e-05,
-      "loss": 0.0895,
+      "loss": 0.1063,
       "step": 3083
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4277762315613695e-05,
-      "loss": 0.0915,
+      "loss": 0.1009,
       "step": 3084
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4275906855923556e-05,
-      "loss": 0.044,
+      "loss": 0.125,
       "step": 3085
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4274051396233417e-05,
-      "loss": 0.0422,
+      "loss": 0.1268,
       "step": 3086
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4272195936543281e-05,
-      "loss": 0.0905,
+      "loss": 0.0988,
       "step": 3087
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4270340476853141e-05,
-      "loss": 0.1916,
+      "loss": 0.0579,
       "step": 3088
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4268485017163003e-05,
-      "loss": 0.1427,
+      "loss": 0.0668,
       "step": 3089
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4266629557472865e-05,
-      "loss": 0.2816,
+      "loss": 0.0725,
       "step": 3090
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4264774097782727e-05,
-      "loss": 0.1875,
+      "loss": 0.0823,
       "step": 3091
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4262918638092589e-05,
-      "loss": 0.0891,
+      "loss": 0.0631,
       "step": 3092
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.426106317840245e-05,
-      "loss": 0.0909,
+      "loss": 0.063,
       "step": 3093
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.425920771871231e-05,
-      "loss": 0.1931,
+      "loss": 0.104,
       "step": 3094
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4257352259022174e-05,
-      "loss": 0.1431,
+      "loss": 0.0246,
       "step": 3095
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4255496799332036e-05,
-      "loss": 0.0888,
+      "loss": 0.0203,
       "step": 3096
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4253641339641896e-05,
-      "loss": 0.1411,
+      "loss": 0.2355,
       "step": 3097
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4251785879951758e-05,
-      "loss": 0.0885,
+      "loss": 0.0623,
       "step": 3098
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4249930420261622e-05,
-      "loss": 0.1955,
+      "loss": 0.1141,
       "step": 3099
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4248074960571484e-05,
-      "loss": 0.1349,
+      "loss": 0.0326,
       "step": 3100
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4246219500881344e-05,
-      "loss": 0.0852,
+      "loss": 0.0236,
       "step": 3101
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4244364041191206e-05,
-      "loss": 0.0362,
+      "loss": 0.0534,
       "step": 3102
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.424250858150107e-05,
-      "loss": 0.1414,
+      "loss": 0.0896,
       "step": 3103
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.424065312181093e-05,
-      "loss": 0.1406,
+      "loss": 0.0522,
       "step": 3104
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4238797662120791e-05,
-      "loss": 0.1392,
+      "loss": 0.1273,
       "step": 3105
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4236942202430653e-05,
-      "loss": 0.184,
+      "loss": 0.0111,
       "step": 3106
     },
     {
       "epoch": 0.86,
       "learning_rate": 1.4235086742740517e-05,
-      "loss": 0.2429,
+      "loss": 0.0255,
       "step": 3107
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4233231283050377e-05,
-      "loss": 0.2528,
+      "loss": 0.0158,
       "step": 3108
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4231375823360239e-05,
-      "loss": 0.1949,
+      "loss": 0.02,
       "step": 3109
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4229520363670099e-05,
-      "loss": 0.3587,
+      "loss": 0.1161,
       "step": 3110
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4227664903979963e-05,
-      "loss": 0.1948,
+      "loss": 0.0506,
       "step": 3111
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4225809444289824e-05,
-      "loss": 0.14,
+      "loss": 0.1055,
       "step": 3112
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4223953984599686e-05,
-      "loss": 0.0895,
+      "loss": 0.0397,
       "step": 3113
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4222098524909546e-05,
-      "loss": 0.24,
+      "loss": 0.1235,
       "step": 3114
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.422024306521941e-05,
-      "loss": 0.0897,
+      "loss": 0.1477,
       "step": 3115
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4218387605529272e-05,
-      "loss": 0.2395,
+      "loss": 0.0137,
       "step": 3116
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4216532145839132e-05,
-      "loss": 0.0888,
+      "loss": 0.0438,
       "step": 3117
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4214676686148994e-05,
-      "loss": 0.0884,
+      "loss": 0.0928,
       "step": 3118
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4212821226458857e-05,
-      "loss": 0.1958,
+      "loss": 0.0926,
       "step": 3119
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4210965766768718e-05,
-      "loss": 0.0877,
+      "loss": 0.0696,
       "step": 3120
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.420911030707858e-05,
-      "loss": 0.1401,
+      "loss": 0.1537,
       "step": 3121
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4207254847388441e-05,
-      "loss": 0.1372,
+      "loss": 0.0514,
       "step": 3122
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4205399387698305e-05,
-      "loss": 0.1855,
+      "loss": 0.1404,
       "step": 3123
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4203543928008165e-05,
-      "loss": 0.088,
+      "loss": 0.0187,
       "step": 3124
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4201688468318027e-05,
-      "loss": 0.2502,
+      "loss": 0.1848,
       "step": 3125
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4199833008627887e-05,
-      "loss": 0.2991,
+      "loss": 0.0117,
       "step": 3126
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.419797754893775e-05,
-      "loss": 0.139,
+      "loss": 0.0958,
       "step": 3127
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4196122089247613e-05,
-      "loss": 0.0889,
+      "loss": 0.0488,
       "step": 3128
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4194266629557474e-05,
-      "loss": 0.1334,
+      "loss": 0.0222,
       "step": 3129
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4192411169867335e-05,
-      "loss": 0.1378,
+      "loss": 0.0186,
       "step": 3130
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4190555710177198e-05,
-      "loss": 0.2986,
+      "loss": 0.1964,
       "step": 3131
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.418870025048706e-05,
-      "loss": 0.1423,
+      "loss": 0.1226,
       "step": 3132
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.418684479079692e-05,
-      "loss": 0.1857,
+      "loss": 0.0215,
       "step": 3133
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4184989331106782e-05,
-      "loss": 0.1887,
+      "loss": 0.0713,
       "step": 3134
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4183133871416646e-05,
-      "loss": 0.1989,
+      "loss": 0.1805,
       "step": 3135
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4181278411726508e-05,
-      "loss": 0.344,
+      "loss": 0.0217,
       "step": 3136
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4179422952036368e-05,
-      "loss": 0.0887,
+      "loss": 0.068,
       "step": 3137
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.417756749234623e-05,
-      "loss": 0.1939,
+      "loss": 0.0368,
       "step": 3138
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4175712032656093e-05,
-      "loss": 0.1383,
+      "loss": 0.0684,
       "step": 3139
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4173856572965953e-05,
-      "loss": 0.0949,
+      "loss": 0.0981,
       "step": 3140
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4172001113275815e-05,
-      "loss": 0.0429,
+      "loss": 0.0325,
       "step": 3141
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4170145653585677e-05,
-      "loss": 0.1937,
+      "loss": 0.0612,
       "step": 3142
     },
     {
       "epoch": 0.87,
       "learning_rate": 1.4168290193895539e-05,
-      "loss": 0.1423,
+      "loss": 0.1016,
       "step": 3143
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.41664347342054e-05,
-      "loss": 0.092,
+      "loss": 0.1218,
       "step": 3144
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4164579274515263e-05,
-      "loss": 0.289,
+      "loss": 0.2063,
       "step": 3145
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4162723814825123e-05,
-      "loss": 0.1408,
+      "loss": 0.0207,
       "step": 3146
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4160868355134986e-05,
-      "loss": 0.0419,
+      "loss": 0.203,
       "step": 3147
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4159012895444848e-05,
-      "loss": 0.1385,
+      "loss": 0.059,
       "step": 3148
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4157157435754709e-05,
-      "loss": 0.135,
+      "loss": 0.1738,
       "step": 3149
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.415530197606457e-05,
-      "loss": 0.0403,
+      "loss": 0.1184,
       "step": 3150
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4153446516374434e-05,
-      "loss": 0.1869,
+      "loss": 0.0242,
       "step": 3151
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4151591056684296e-05,
-      "loss": 0.09,
+      "loss": 0.0211,
       "step": 3152
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4149735596994156e-05,
-      "loss": 0.1399,
+      "loss": 0.048,
       "step": 3153
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4147880137304018e-05,
-      "loss": 0.1387,
+      "loss": 0.0274,
       "step": 3154
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4146024677613881e-05,
-      "loss": 0.2941,
+      "loss": 0.0689,
       "step": 3155
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4144169217923742e-05,
-      "loss": 0.142,
+      "loss": 0.1136,
       "step": 3156
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4142313758233603e-05,
-      "loss": 0.089,
+      "loss": 0.061,
       "step": 3157
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4140458298543465e-05,
-      "loss": 0.1407,
+      "loss": 0.2634,
       "step": 3158
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4138602838853329e-05,
-      "loss": 0.2431,
+      "loss": 0.0191,
       "step": 3159
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4136747379163189e-05,
-      "loss": 0.246,
+      "loss": 0.2015,
       "step": 3160
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4134891919473051e-05,
-      "loss": 0.1953,
+      "loss": 0.0179,
       "step": 3161
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4133036459782911e-05,
-      "loss": 0.0883,
+      "loss": 0.0195,
       "step": 3162
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4131181000092775e-05,
-      "loss": 0.0837,
+      "loss": 0.0508,
       "step": 3163
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4129325540402637e-05,
-      "loss": 0.0808,
+      "loss": 0.0199,
       "step": 3164
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4127470080712498e-05,
-      "loss": 0.1926,
+      "loss": 0.0804,
       "step": 3165
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4125614621022359e-05,
-      "loss": 0.1401,
+      "loss": 0.0654,
       "step": 3166
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4123759161332222e-05,
-      "loss": 0.1901,
+      "loss": 0.1657,
       "step": 3167
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4121903701642084e-05,
-      "loss": 0.3551,
+      "loss": 0.1358,
       "step": 3168
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4120048241951944e-05,
-      "loss": 0.085,
+      "loss": 0.0617,
       "step": 3169
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4118192782261806e-05,
-      "loss": 0.2414,
+      "loss": 0.0594,
       "step": 3170
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.411633732257167e-05,
-      "loss": 0.0371,
+      "loss": 0.1108,
       "step": 3171
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.411448186288153e-05,
-      "loss": 0.0875,
+      "loss": 0.0773,
       "step": 3172
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4112626403191392e-05,
-      "loss": 0.0368,
+      "loss": 0.0506,
       "step": 3173
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4110770943501254e-05,
-      "loss": 0.1872,
+      "loss": 0.0635,
       "step": 3174
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4108915483811117e-05,
-      "loss": 0.0922,
+      "loss": 0.0517,
       "step": 3175
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4107060024120977e-05,
-      "loss": 0.1937,
+      "loss": 0.0885,
       "step": 3176
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.410520456443084e-05,
-      "loss": 0.0348,
+      "loss": 0.0195,
       "step": 3177
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.41033491047407e-05,
-      "loss": 0.191,
+      "loss": 0.1001,
       "step": 3178
     },
     {
       "epoch": 0.88,
       "learning_rate": 1.4101493645050563e-05,
-      "loss": 0.1383,
+      "loss": 0.0837,
       "step": 3179
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4099638185360425e-05,
-      "loss": 0.19,
+      "loss": 0.0112,
       "step": 3180
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4097782725670287e-05,
-      "loss": 0.2944,
+      "loss": 0.1106,
       "step": 3181
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4095927265980147e-05,
-      "loss": 0.086,
+      "loss": 0.0361,
       "step": 3182
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.409407180629001e-05,
-      "loss": 0.2406,
+      "loss": 0.1781,
       "step": 3183
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4092216346599872e-05,
-      "loss": 0.138,
+      "loss": 0.1238,
       "step": 3184
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4090360886909732e-05,
-      "loss": 0.1362,
+      "loss": 0.129,
       "step": 3185
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4088505427219594e-05,
-      "loss": 0.2991,
+      "loss": 0.0616,
       "step": 3186
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4086649967529458e-05,
-      "loss": 0.1381,
+      "loss": 0.0556,
       "step": 3187
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4084794507839318e-05,
-      "loss": 0.142,
+      "loss": 0.073,
       "step": 3188
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.408293904814918e-05,
-      "loss": 0.1936,
+      "loss": 0.0142,
       "step": 3189
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4081083588459042e-05,
-      "loss": 0.1893,
+      "loss": 0.0994,
       "step": 3190
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4079228128768902e-05,
-      "loss": 0.1414,
+      "loss": 0.0668,
       "step": 3191
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4077372669078766e-05,
-      "loss": 0.2483,
+      "loss": 0.0541,
       "step": 3192
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4075517209388627e-05,
-      "loss": 0.189,
+      "loss": 0.1254,
       "step": 3193
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4073661749698488e-05,
-      "loss": 0.1368,
+      "loss": 0.0827,
       "step": 3194
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.407180629000835e-05,
-      "loss": 0.1918,
+      "loss": 0.0482,
       "step": 3195
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4069950830318213e-05,
-      "loss": 0.0882,
+      "loss": 0.0239,
       "step": 3196
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4068095370628075e-05,
-      "loss": 0.1898,
+      "loss": 0.118,
       "step": 3197
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4066239910937935e-05,
-      "loss": 0.3908,
+      "loss": 0.1409,
       "step": 3198
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4064384451247797e-05,
-      "loss": 0.1375,
+      "loss": 0.0501,
       "step": 3199
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.406252899155766e-05,
-      "loss": 0.1894,
+      "loss": 0.1427,
       "step": 3200
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.406067353186752e-05,
-      "loss": 0.0448,
+      "loss": 0.0721,
       "step": 3201
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4058818072177383e-05,
-      "loss": 0.0918,
+      "loss": 0.1672,
       "step": 3202
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4056962612487244e-05,
-      "loss": 0.0929,
+      "loss": 0.0277,
       "step": 3203
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4055107152797108e-05,
-      "loss": 0.1877,
+      "loss": 0.0868,
       "step": 3204
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4053251693106968e-05,
-      "loss": 0.2373,
+      "loss": 0.1297,
       "step": 3205
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.405139623341683e-05,
-      "loss": 0.1865,
+      "loss": 0.0988,
       "step": 3206
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.404954077372669e-05,
-      "loss": 0.186,
+      "loss": 0.0891,
       "step": 3207
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4047685314036554e-05,
-      "loss": 0.1454,
+      "loss": 0.0623,
       "step": 3208
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4045829854346416e-05,
-      "loss": 0.091,
+      "loss": 0.0268,
       "step": 3209
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4043974394656278e-05,
-      "loss": 0.1412,
+      "loss": 0.1342,
       "step": 3210
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4042118934966138e-05,
-      "loss": 0.2436,
+      "loss": 0.0651,
       "step": 3211
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4040263475276001e-05,
-      "loss": 0.2936,
+      "loss": 0.0317,
       "step": 3212
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4038408015585863e-05,
-      "loss": 0.1383,
+      "loss": 0.0189,
       "step": 3213
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4036552555895723e-05,
-      "loss": 0.0414,
+      "loss": 0.0606,
       "step": 3214
     },
     {
       "epoch": 0.89,
       "learning_rate": 1.4034697096205585e-05,
-      "loss": 0.1923,
+      "loss": 0.019,
       "step": 3215
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4032841636515449e-05,
-      "loss": 0.2855,
+      "loss": 0.0551,
       "step": 3216
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4030986176825309e-05,
-      "loss": 0.0862,
+      "loss": 0.1807,
       "step": 3217
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.402913071713517e-05,
-      "loss": 0.1854,
+      "loss": 0.0698,
       "step": 3218
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4027275257445033e-05,
-      "loss": 0.1385,
+      "loss": 0.0682,
       "step": 3219
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4025419797754896e-05,
-      "loss": 0.2433,
+      "loss": 0.0353,
       "step": 3220
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4023564338064756e-05,
-      "loss": 0.1892,
+      "loss": 0.1306,
       "step": 3221
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4021708878374618e-05,
-      "loss": 0.1897,
+      "loss": 0.0794,
       "step": 3222
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4019853418684478e-05,
-      "loss": 0.1345,
+      "loss": 0.0116,
       "step": 3223
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4017997958994342e-05,
-      "loss": 0.238,
+      "loss": 0.0738,
       "step": 3224
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4016142499304204e-05,
-      "loss": 0.1872,
+      "loss": 0.159,
       "step": 3225
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4014287039614066e-05,
-      "loss": 0.1897,
+      "loss": 0.2471,
       "step": 3226
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4012431579923926e-05,
-      "loss": 0.0889,
+      "loss": 0.0938,
       "step": 3227
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.401057612023379e-05,
-      "loss": 0.2373,
+      "loss": 0.1635,
       "step": 3228
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4008720660543651e-05,
-      "loss": 0.0945,
+      "loss": 0.0431,
       "step": 3229
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4006865200853512e-05,
-      "loss": 0.0453,
+      "loss": 0.0892,
       "step": 3230
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4005009741163373e-05,
-      "loss": 0.1432,
+      "loss": 0.0294,
       "step": 3231
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4003154281473237e-05,
-      "loss": 0.284,
+      "loss": 0.082,
       "step": 3232
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.4001298821783099e-05,
-      "loss": 0.187,
+      "loss": 0.1238,
       "step": 3233
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3999443362092959e-05,
-      "loss": 0.1872,
+      "loss": 0.0296,
       "step": 3234
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3997587902402821e-05,
-      "loss": 0.339,
+      "loss": 0.0715,
       "step": 3235
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3995732442712684e-05,
-      "loss": 0.3292,
+      "loss": 0.0301,
       "step": 3236
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3993876983022545e-05,
-      "loss": 0.2357,
+      "loss": 0.0739,
       "step": 3237
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3992021523332407e-05,
-      "loss": 0.0946,
+      "loss": 0.0503,
       "step": 3238
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3990166063642268e-05,
-      "loss": 0.1375,
+      "loss": 0.023,
       "step": 3239
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.398831060395213e-05,
-      "loss": 0.0481,
+      "loss": 0.1101,
       "step": 3240
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3986455144261992e-05,
-      "loss": 0.2289,
+      "loss": 0.0674,
       "step": 3241
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3984599684571854e-05,
-      "loss": 0.0988,
+      "loss": 0.0549,
       "step": 3242
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3982744224881714e-05,
-      "loss": 0.0964,
+      "loss": 0.024,
       "step": 3243
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3980888765191578e-05,
-      "loss": 0.0951,
+      "loss": 0.0937,
       "step": 3244
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.397903330550144e-05,
-      "loss": 0.3858,
+      "loss": 0.0171,
       "step": 3245
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.39771778458113e-05,
-      "loss": 0.2351,
+      "loss": 0.0568,
       "step": 3246
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3975322386121162e-05,
-      "loss": 0.0447,
+      "loss": 0.1426,
       "step": 3247
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3973466926431025e-05,
-      "loss": 0.1904,
+      "loss": 0.0162,
       "step": 3248
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3971611466740887e-05,
-      "loss": 0.0944,
+      "loss": 0.0922,
       "step": 3249
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3969756007050747e-05,
-      "loss": 0.1462,
+      "loss": 0.104,
       "step": 3250
     },
     {
       "epoch": 0.9,
       "learning_rate": 1.3967900547360609e-05,
-      "loss": 0.2892,
+      "loss": 0.202,
       "step": 3251
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3966045087670473e-05,
-      "loss": 0.0897,
+      "loss": 0.0883,
       "step": 3252
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3964189627980333e-05,
-      "loss": 0.2381,
+      "loss": 0.0813,
       "step": 3253
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3962334168290195e-05,
-      "loss": 0.1404,
+      "loss": 0.0664,
       "step": 3254
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3960478708600057e-05,
-      "loss": 0.1863,
+      "loss": 0.0701,
       "step": 3255
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.395862324890992e-05,
-      "loss": 0.0935,
+      "loss": 0.1106,
       "step": 3256
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.395676778921978e-05,
-      "loss": 0.1943,
+      "loss": 0.1297,
       "step": 3257
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3954912329529642e-05,
-      "loss": 0.1408,
+      "loss": 0.0569,
       "step": 3258
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3953056869839502e-05,
-      "loss": 0.0915,
+      "loss": 0.1619,
       "step": 3259
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3951201410149366e-05,
-      "loss": 0.1402,
+      "loss": 0.2024,
       "step": 3260
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3949345950459228e-05,
-      "loss": 0.0386,
+      "loss": 0.0261,
       "step": 3261
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.394749049076909e-05,
-      "loss": 0.0893,
+      "loss": 0.108,
       "step": 3262
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.394563503107895e-05,
-      "loss": 0.0901,
+      "loss": 0.0887,
       "step": 3263
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3943779571388813e-05,
-      "loss": 0.1903,
+      "loss": 0.0239,
       "step": 3264
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3941924111698675e-05,
-      "loss": 0.1408,
+      "loss": 0.029,
       "step": 3265
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3940068652008535e-05,
-      "loss": 0.3468,
+      "loss": 0.036,
       "step": 3266
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3938213192318397e-05,
-      "loss": 0.2396,
+      "loss": 0.0456,
       "step": 3267
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3936357732628261e-05,
-      "loss": 0.1847,
+      "loss": 0.0256,
       "step": 3268
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3934502272938121e-05,
-      "loss": 0.0867,
+      "loss": 0.1113,
       "step": 3269
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3932646813247983e-05,
-      "loss": 0.2413,
+      "loss": 0.0713,
       "step": 3270
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3930791353557845e-05,
-      "loss": 0.1395,
+      "loss": 0.0429,
       "step": 3271
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3928935893867708e-05,
-      "loss": 0.2402,
+      "loss": 0.2337,
       "step": 3272
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3927080434177569e-05,
-      "loss": 0.0878,
+      "loss": 0.1197,
       "step": 3273
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.392522497448743e-05,
-      "loss": 0.1418,
+      "loss": 0.0649,
       "step": 3274
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.392336951479729e-05,
-      "loss": 0.2451,
+      "loss": 0.0164,
       "step": 3275
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3921514055107154e-05,
-      "loss": 0.2374,
+      "loss": 0.058,
       "step": 3276
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3919658595417016e-05,
-      "loss": 0.1383,
+      "loss": 0.0225,
       "step": 3277
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3917803135726878e-05,
-      "loss": 0.1395,
+      "loss": 0.0137,
       "step": 3278
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3915947676036738e-05,
-      "loss": 0.0385,
+      "loss": 0.0996,
       "step": 3279
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3914092216346602e-05,
-      "loss": 0.1408,
+      "loss": 0.013,
       "step": 3280
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3912236756656464e-05,
-      "loss": 0.0919,
+      "loss": 0.2657,
       "step": 3281
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3910381296966324e-05,
-      "loss": 0.1377,
+      "loss": 0.3283,
       "step": 3282
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3908525837276186e-05,
-      "loss": 0.0883,
+      "loss": 0.1002,
       "step": 3283
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.390667037758605e-05,
-      "loss": 0.2414,
+      "loss": 0.1301,
       "step": 3284
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3904814917895911e-05,
-      "loss": 0.2948,
+      "loss": 0.1011,
       "step": 3285
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3902959458205771e-05,
-      "loss": 0.0866,
+      "loss": 0.0583,
       "step": 3286
     },
     {
       "epoch": 0.91,
       "learning_rate": 1.3901103998515633e-05,
-      "loss": 0.087,
+      "loss": 0.1008,
       "step": 3287
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3899248538825497e-05,
-      "loss": 0.2413,
+      "loss": 0.0993,
       "step": 3288
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3897393079135357e-05,
-      "loss": 0.0373,
+      "loss": 0.0216,
       "step": 3289
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3895537619445219e-05,
-      "loss": 0.088,
+      "loss": 0.0673,
       "step": 3290
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.389368215975508e-05,
-      "loss": 0.1358,
+      "loss": 0.0715,
       "step": 3291
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3891826700064942e-05,
-      "loss": 0.4028,
+      "loss": 0.0245,
       "step": 3292
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3889971240374804e-05,
-      "loss": 0.1413,
+      "loss": 0.025,
       "step": 3293
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3888115780684666e-05,
-      "loss": 0.235,
+      "loss": 0.0793,
       "step": 3294
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3886260320994526e-05,
-      "loss": 0.0878,
+      "loss": 0.0438,
       "step": 3295
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.388440486130439e-05,
-      "loss": 0.346,
+      "loss": 0.0208,
       "step": 3296
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3882549401614252e-05,
-      "loss": 0.139,
+      "loss": 0.0153,
       "step": 3297
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3880693941924112e-05,
-      "loss": 0.1942,
+      "loss": 0.0647,
       "step": 3298
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3878838482233974e-05,
-      "loss": 0.2438,
+      "loss": 0.0209,
       "step": 3299
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3876983022543837e-05,
-      "loss": 0.0898,
+      "loss": 0.06,
       "step": 3300
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.38751275628537e-05,
-      "loss": 0.0917,
+      "loss": 0.1171,
       "step": 3301
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.387327210316356e-05,
-      "loss": 0.2395,
+      "loss": 0.0666,
       "step": 3302
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3871416643473421e-05,
-      "loss": 0.1384,
+      "loss": 0.1569,
       "step": 3303
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3869561183783285e-05,
-      "loss": 0.1437,
+      "loss": 0.0127,
       "step": 3304
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3867705724093145e-05,
-      "loss": 0.1392,
+      "loss": 0.0099,
       "step": 3305
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3865850264403007e-05,
-      "loss": 0.0884,
+      "loss": 0.013,
       "step": 3306
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3863994804712869e-05,
-      "loss": 0.1927,
+      "loss": 0.1594,
       "step": 3307
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.386213934502273e-05,
-      "loss": 0.1918,
+      "loss": 0.1088,
       "step": 3308
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3860283885332593e-05,
-      "loss": 0.0387,
+      "loss": 0.0534,
       "step": 3309
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3858428425642454e-05,
-      "loss": 0.141,
+      "loss": 0.1987,
       "step": 3310
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3856572965952315e-05,
-      "loss": 0.2924,
+      "loss": 0.119,
       "step": 3311
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3854717506262178e-05,
-      "loss": 0.1354,
+      "loss": 0.1136,
       "step": 3312
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.385286204657204e-05,
-      "loss": 0.1429,
+      "loss": 0.0599,
       "step": 3313
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.38510065868819e-05,
-      "loss": 0.1908,
+      "loss": 0.2213,
       "step": 3314
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3849151127191762e-05,
-      "loss": 0.0403,
+      "loss": 0.1,
       "step": 3315
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3847295667501626e-05,
-      "loss": 0.1903,
+      "loss": 0.0242,
       "step": 3316
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3845440207811488e-05,
-      "loss": 0.0923,
+      "loss": 0.0694,
       "step": 3317
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3843584748121348e-05,
-      "loss": 0.0388,
+      "loss": 0.0596,
       "step": 3318
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.384172928843121e-05,
-      "loss": 0.1938,
+      "loss": 0.0878,
       "step": 3319
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3839873828741073e-05,
-      "loss": 0.0374,
+      "loss": 0.0286,
       "step": 3320
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3838018369050933e-05,
-      "loss": 0.0861,
+      "loss": 0.1957,
       "step": 3321
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3836162909360795e-05,
-      "loss": 0.1377,
+      "loss": 0.02,
       "step": 3322
     },
     {
       "epoch": 0.92,
       "learning_rate": 1.3834307449670657e-05,
-      "loss": 0.2479,
+      "loss": 0.0862,
       "step": 3323
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.383245198998052e-05,
-      "loss": 0.0871,
+      "loss": 0.1362,
       "step": 3324
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.383059653029038e-05,
-      "loss": 0.0869,
+      "loss": 0.0224,
       "step": 3325
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3828741070600243e-05,
-      "loss": 0.0853,
+      "loss": 0.0559,
       "step": 3326
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3826885610910103e-05,
-      "loss": 0.2427,
+      "loss": 0.0609,
       "step": 3327
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3825030151219966e-05,
-      "loss": 0.1438,
+      "loss": 0.0553,
       "step": 3328
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3823174691529828e-05,
-      "loss": 0.0319,
+      "loss": 0.1048,
       "step": 3329
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.382131923183969e-05,
-      "loss": 0.0311,
+      "loss": 0.209,
       "step": 3330
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.381946377214955e-05,
-      "loss": 0.1961,
+      "loss": 0.0997,
       "step": 3331
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3817608312459414e-05,
-      "loss": 0.3567,
+      "loss": 0.1355,
       "step": 3332
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3815752852769276e-05,
-      "loss": 0.0858,
+      "loss": 0.0615,
       "step": 3333
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3813897393079136e-05,
-      "loss": 0.086,
+      "loss": 0.1038,
       "step": 3334
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3812041933388998e-05,
-      "loss": 0.1426,
+      "loss": 0.1204,
       "step": 3335
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3810186473698861e-05,
-      "loss": 0.1426,
+      "loss": 0.0807,
       "step": 3336
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3808331014008722e-05,
-      "loss": 0.3072,
+      "loss": 0.2308,
       "step": 3337
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3806475554318583e-05,
-      "loss": 0.195,
+      "loss": 0.1033,
       "step": 3338
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3804620094628445e-05,
-      "loss": 0.0302,
+      "loss": 0.0938,
       "step": 3339
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3802764634938305e-05,
-      "loss": 0.1419,
+      "loss": 0.0644,
       "step": 3340
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3800909175248169e-05,
-      "loss": 0.0844,
+      "loss": 0.0618,
       "step": 3341
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3799053715558031e-05,
-      "loss": 0.3003,
+      "loss": 0.0919,
       "step": 3342
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3797198255867891e-05,
-      "loss": 0.0859,
+      "loss": 0.0491,
       "step": 3343
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3795342796177753e-05,
-      "loss": 0.0318,
+      "loss": 0.0568,
       "step": 3344
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3793487336487616e-05,
-      "loss": 0.3016,
+      "loss": 0.0505,
       "step": 3345
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3791631876797478e-05,
-      "loss": 0.1921,
+      "loss": 0.0896,
       "step": 3346
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3789776417107339e-05,
-      "loss": 0.194,
+      "loss": 0.0983,
       "step": 3347
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.37879209574172e-05,
-      "loss": 0.0853,
+      "loss": 0.0428,
       "step": 3348
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3786065497727064e-05,
-      "loss": 0.1417,
+      "loss": 0.2701,
       "step": 3349
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3784210038036924e-05,
-      "loss": 0.2429,
+      "loss": 0.2988,
       "step": 3350
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3782354578346786e-05,
-      "loss": 0.3014,
+      "loss": 0.0535,
       "step": 3351
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3780499118656648e-05,
-      "loss": 0.1873,
+      "loss": 0.1732,
       "step": 3352
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3778643658966511e-05,
-      "loss": 0.1392,
+      "loss": 0.1498,
       "step": 3353
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3776788199276372e-05,
-      "loss": 0.2991,
+      "loss": 0.0263,
       "step": 3354
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3774932739586234e-05,
-      "loss": 0.0873,
+      "loss": 0.1087,
       "step": 3355
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3773077279896094e-05,
-      "loss": 0.1896,
+      "loss": 0.0559,
       "step": 3356
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3771221820205957e-05,
-      "loss": 0.0896,
+      "loss": 0.073,
       "step": 3357
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3769366360515819e-05,
-      "loss": 0.1902,
+      "loss": 0.0879,
       "step": 3358
     },
     {
       "epoch": 0.93,
       "learning_rate": 1.3767510900825681e-05,
-      "loss": 0.2353,
+      "loss": 0.061,
       "step": 3359
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3765655441135541e-05,
-      "loss": 0.1324,
+      "loss": 0.1546,
       "step": 3360
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3763799981445405e-05,
-      "loss": 0.1928,
+      "loss": 0.1254,
       "step": 3361
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3761944521755267e-05,
-      "loss": 0.0949,
+      "loss": 0.0295,
       "step": 3362
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3760089062065127e-05,
-      "loss": 0.1427,
+      "loss": 0.0497,
       "step": 3363
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3758233602374989e-05,
-      "loss": 0.1873,
+      "loss": 0.0594,
       "step": 3364
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3756378142684852e-05,
-      "loss": 0.2811,
+      "loss": 0.0269,
       "step": 3365
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3754522682994712e-05,
-      "loss": 0.1455,
+      "loss": 0.0667,
       "step": 3366
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3752667223304574e-05,
-      "loss": 0.0947,
+      "loss": 0.0234,
       "step": 3367
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3750811763614436e-05,
-      "loss": 0.0903,
+      "loss": 0.1025,
       "step": 3368
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.37489563039243e-05,
-      "loss": 0.046,
+      "loss": 0.0203,
       "step": 3369
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.374710084423416e-05,
-      "loss": 0.2396,
+      "loss": 0.1105,
       "step": 3370
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3745245384544022e-05,
-      "loss": 0.0895,
+      "loss": 0.1616,
       "step": 3371
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3743389924853882e-05,
-      "loss": 0.0421,
+      "loss": 0.0631,
       "step": 3372
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3741534465163745e-05,
-      "loss": 0.1395,
+      "loss": 0.1521,
       "step": 3373
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3739679005473607e-05,
-      "loss": 0.1902,
+      "loss": 0.0173,
       "step": 3374
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.373782354578347e-05,
-      "loss": 0.1411,
+      "loss": 0.0642,
       "step": 3375
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.373596808609333e-05,
-      "loss": 0.2926,
+      "loss": 0.1252,
       "step": 3376
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3734112626403193e-05,
-      "loss": 0.0373,
+      "loss": 0.0613,
       "step": 3377
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3732257166713055e-05,
-      "loss": 0.2455,
+      "loss": 0.0935,
       "step": 3378
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3730401707022915e-05,
-      "loss": 0.1893,
+      "loss": 0.0144,
       "step": 3379
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3728546247332777e-05,
-      "loss": 0.0892,
+      "loss": 0.0699,
       "step": 3380
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.372669078764264e-05,
-      "loss": 0.0358,
+      "loss": 0.0177,
       "step": 3381
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3724835327952502e-05,
-      "loss": 0.1398,
+      "loss": 0.1725,
       "step": 3382
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3722979868262362e-05,
-      "loss": 0.1424,
+      "loss": 0.0939,
       "step": 3383
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3721124408572224e-05,
-      "loss": 0.0858,
+      "loss": 0.0382,
       "step": 3384
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3719268948882088e-05,
-      "loss": 0.1416,
+      "loss": 0.119,
       "step": 3385
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3717413489191948e-05,
-      "loss": 0.139,
+      "loss": 0.0584,
       "step": 3386
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.371555802950181e-05,
-      "loss": 0.0844,
+      "loss": 0.0364,
       "step": 3387
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3713702569811672e-05,
-      "loss": 0.0846,
+      "loss": 0.0241,
       "step": 3388
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3711847110121534e-05,
-      "loss": 0.1994,
+      "loss": 0.0502,
       "step": 3389
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3709991650431396e-05,
-      "loss": 0.1359,
+      "loss": 0.0579,
       "step": 3390
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3708136190741257e-05,
-      "loss": 0.242,
+      "loss": 0.1322,
       "step": 3391
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3706280731051118e-05,
-      "loss": 0.1948,
+      "loss": 0.0634,
       "step": 3392
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3704425271360981e-05,
-      "loss": 0.082,
+      "loss": 0.0646,
       "step": 3393
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3702569811670843e-05,
-      "loss": 0.0307,
+      "loss": 0.0538,
       "step": 3394
     },
     {
       "epoch": 0.94,
       "learning_rate": 1.3700714351980703e-05,
-      "loss": 0.0848,
+      "loss": 0.0211,
       "step": 3395
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3698858892290565e-05,
-      "loss": 0.0869,
+      "loss": 0.0614,
       "step": 3396
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3697003432600429e-05,
-      "loss": 0.2449,
+      "loss": 0.0874,
       "step": 3397
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.369514797291029e-05,
-      "loss": 0.0858,
+      "loss": 0.1349,
       "step": 3398
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.369329251322015e-05,
-      "loss": 0.138,
+      "loss": 0.0148,
       "step": 3399
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3691437053530013e-05,
-      "loss": 0.1909,
+      "loss": 0.0859,
       "step": 3400
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3689581593839876e-05,
-      "loss": 0.1912,
+      "loss": 0.1305,
       "step": 3401
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3687726134149736e-05,
-      "loss": 0.1406,
+      "loss": 0.0572,
       "step": 3402
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3685870674459598e-05,
-      "loss": 0.0826,
+      "loss": 0.1113,
       "step": 3403
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.368401521476946e-05,
-      "loss": 0.1382,
+      "loss": 0.0339,
       "step": 3404
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3682159755079324e-05,
-      "loss": 0.251,
+      "loss": 0.1702,
       "step": 3405
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3680304295389184e-05,
-      "loss": 0.2499,
+      "loss": 0.0596,
       "step": 3406
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3678448835699046e-05,
-      "loss": 0.2451,
+      "loss": 0.0581,
       "step": 3407
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3676593376008906e-05,
-      "loss": 0.0869,
+      "loss": 0.016,
       "step": 3408
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.367473791631877e-05,
-      "loss": 0.3028,
+      "loss": 0.0744,
       "step": 3409
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3672882456628631e-05,
-      "loss": 0.0834,
+      "loss": 0.0669,
       "step": 3410
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3671026996938493e-05,
-      "loss": 0.1399,
+      "loss": 0.06,
       "step": 3411
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3669171537248353e-05,
-      "loss": 0.1916,
+      "loss": 0.0483,
       "step": 3412
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3667316077558217e-05,
-      "loss": 0.0337,
+      "loss": 0.0103,
       "step": 3413
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3665460617868079e-05,
-      "loss": 0.246,
+      "loss": 0.098,
       "step": 3414
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3663605158177939e-05,
-      "loss": 0.1391,
+      "loss": 0.0542,
       "step": 3415
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.36617496984878e-05,
-      "loss": 0.0884,
+      "loss": 0.0654,
       "step": 3416
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3659894238797664e-05,
-      "loss": 0.1416,
+      "loss": 0.0991,
       "step": 3417
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3658038779107525e-05,
-      "loss": 0.2965,
+      "loss": 0.1221,
       "step": 3418
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3656183319417386e-05,
-      "loss": 0.1434,
+      "loss": 0.0484,
       "step": 3419
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3654327859727248e-05,
-      "loss": 0.1385,
+      "loss": 0.1649,
       "step": 3420
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3652472400037112e-05,
-      "loss": 0.2408,
+      "loss": 0.1003,
       "step": 3421
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3650616940346972e-05,
-      "loss": 0.1852,
+      "loss": 0.0141,
       "step": 3422
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3648761480656834e-05,
-      "loss": 0.0396,
+      "loss": 0.0497,
       "step": 3423
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3646906020966694e-05,
-      "loss": 0.0911,
+      "loss": 0.0158,
       "step": 3424
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3645050561276558e-05,
-      "loss": 0.1923,
+      "loss": 0.0125,
       "step": 3425
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.364319510158642e-05,
-      "loss": 0.0874,
+      "loss": 0.0517,
       "step": 3426
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3641339641896281e-05,
-      "loss": 0.0377,
+      "loss": 0.0303,
       "step": 3427
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3639484182206142e-05,
-      "loss": 0.0903,
+      "loss": 0.0194,
       "step": 3428
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3637628722516005e-05,
-      "loss": 0.1375,
+      "loss": 0.0692,
       "step": 3429
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3635773262825867e-05,
-      "loss": 0.2442,
+      "loss": 0.0125,
       "step": 3430
     },
     {
       "epoch": 0.95,
       "learning_rate": 1.3633917803135727e-05,
-      "loss": 0.0361,
+      "loss": 0.056,
       "step": 3431
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3632062343445589e-05,
-      "loss": 0.1906,
+      "loss": 0.2221,
       "step": 3432
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3630206883755453e-05,
-      "loss": 0.1384,
+      "loss": 0.1815,
       "step": 3433
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3628351424065313e-05,
-      "loss": 0.1428,
+      "loss": 0.0575,
       "step": 3434
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3626495964375175e-05,
-      "loss": 0.0847,
+      "loss": 0.1275,
       "step": 3435
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3624640504685037e-05,
-      "loss": 0.1422,
+      "loss": 0.2304,
       "step": 3436
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.36227850449949e-05,
-      "loss": 0.139,
+      "loss": 0.0539,
       "step": 3437
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.362092958530476e-05,
-      "loss": 0.1934,
+      "loss": 0.0532,
       "step": 3438
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3619074125614622e-05,
-      "loss": 0.239,
+      "loss": 0.0516,
       "step": 3439
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3617218665924482e-05,
-      "loss": 0.1836,
+      "loss": 0.0192,
       "step": 3440
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3615363206234346e-05,
-      "loss": 0.3534,
+      "loss": 0.0213,
       "step": 3441
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3613507746544208e-05,
-      "loss": 0.1379,
+      "loss": 0.1157,
       "step": 3442
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.361165228685407e-05,
-      "loss": 0.1859,
+      "loss": 0.0221,
       "step": 3443
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.360979682716393e-05,
-      "loss": 0.3026,
+      "loss": 0.1822,
       "step": 3444
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3607941367473793e-05,
-      "loss": 0.359,
+      "loss": 0.0714,
       "step": 3445
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3606085907783655e-05,
-      "loss": 0.1393,
+      "loss": 0.0785,
       "step": 3446
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3604230448093515e-05,
-      "loss": 0.0392,
+      "loss": 0.0882,
       "step": 3447
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3602374988403377e-05,
-      "loss": 0.1456,
+      "loss": 0.1014,
       "step": 3448
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3600519528713241e-05,
-      "loss": 0.1891,
+      "loss": 0.0208,
       "step": 3449
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3598664069023103e-05,
-      "loss": 0.1902,
+      "loss": 0.0248,
       "step": 3450
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3596808609332963e-05,
-      "loss": 0.1393,
+      "loss": 0.101,
       "step": 3451
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3594953149642825e-05,
-      "loss": 0.0803,
+      "loss": 0.0309,
       "step": 3452
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3593097689952688e-05,
-      "loss": 0.1001,
+      "loss": 0.0605,
       "step": 3453
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3591242230262549e-05,
-      "loss": 0.1968,
+      "loss": 0.0901,
       "step": 3454
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.358938677057241e-05,
-      "loss": 0.2399,
+      "loss": 0.0896,
       "step": 3455
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3587531310882272e-05,
-      "loss": 0.1909,
+      "loss": 0.0694,
       "step": 3456
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3585675851192134e-05,
-      "loss": 0.1411,
+      "loss": 0.1677,
       "step": 3457
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3583820391501996e-05,
-      "loss": 0.0922,
+      "loss": 0.0524,
       "step": 3458
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3581964931811858e-05,
-      "loss": 0.1929,
+      "loss": 0.0658,
       "step": 3459
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3580109472121718e-05,
-      "loss": 0.0965,
+      "loss": 0.1023,
       "step": 3460
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3578254012431582e-05,
-      "loss": 0.0426,
+      "loss": 0.1023,
       "step": 3461
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3576398552741443e-05,
-      "loss": 0.0931,
+      "loss": 0.0253,
       "step": 3462
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3574543093051304e-05,
-      "loss": 0.1967,
+      "loss": 0.0332,
       "step": 3463
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3572687633361166e-05,
-      "loss": 0.1899,
+      "loss": 0.0268,
       "step": 3464
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3570832173671029e-05,
-      "loss": 0.0375,
+      "loss": 0.0655,
       "step": 3465
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3568976713980891e-05,
-      "loss": 0.0871,
+      "loss": 0.0936,
       "step": 3466
     },
     {
       "epoch": 0.96,
       "learning_rate": 1.3567121254290751e-05,
-      "loss": 0.141,
+      "loss": 0.0253,
       "step": 3467
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3565265794600613e-05,
-      "loss": 0.1936,
+      "loss": 0.1215,
       "step": 3468
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3563410334910477e-05,
-      "loss": 0.244,
+      "loss": 0.0603,
       "step": 3469
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3561554875220337e-05,
-      "loss": 0.1386,
+      "loss": 0.0773,
       "step": 3470
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3559699415530199e-05,
-      "loss": 0.1956,
+      "loss": 0.0614,
       "step": 3471
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.355784395584006e-05,
-      "loss": 0.0334,
+      "loss": 0.0988,
       "step": 3472
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3555988496149924e-05,
-      "loss": 0.139,
+      "loss": 0.1275,
       "step": 3473
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3554133036459784e-05,
-      "loss": 0.3525,
+      "loss": 0.0663,
       "step": 3474
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3552277576769646e-05,
-      "loss": 0.0349,
+      "loss": 0.1246,
       "step": 3475
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3550422117079506e-05,
-      "loss": 0.1413,
+      "loss": 0.102,
       "step": 3476
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.354856665738937e-05,
-      "loss": 0.142,
+      "loss": 0.0618,
       "step": 3477
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3546711197699232e-05,
-      "loss": 0.1411,
+      "loss": 0.1284,
       "step": 3478
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3544855738009094e-05,
-      "loss": 0.1381,
+      "loss": 0.015,
       "step": 3479
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3543000278318954e-05,
-      "loss": 0.2516,
+      "loss": 0.0191,
       "step": 3480
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3541144818628817e-05,
-      "loss": 0.287,
+      "loss": 0.1104,
       "step": 3481
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.353928935893868e-05,
-      "loss": 0.0331,
+      "loss": 0.0601,
       "step": 3482
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.353743389924854e-05,
-      "loss": 0.0816,
+      "loss": 0.1219,
       "step": 3483
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3535578439558401e-05,
-      "loss": 0.1385,
+      "loss": 0.2037,
       "step": 3484
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3533722979868265e-05,
-      "loss": 0.0328,
+      "loss": 0.0222,
       "step": 3485
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3531867520178125e-05,
-      "loss": 0.2485,
+      "loss": 0.0189,
       "step": 3486
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3530012060487987e-05,
-      "loss": 0.0844,
+      "loss": 0.0418,
       "step": 3487
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3528156600797849e-05,
-      "loss": 0.1958,
+      "loss": 0.1183,
       "step": 3488
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3526301141107712e-05,
-      "loss": 0.087,
+      "loss": 0.1521,
       "step": 3489
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3524445681417572e-05,
-      "loss": 0.1386,
+      "loss": 0.0109,
       "step": 3490
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3522590221727434e-05,
-      "loss": 0.1929,
+      "loss": 0.0588,
       "step": 3491
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3520734762037295e-05,
-      "loss": 0.1386,
+      "loss": 0.0432,
       "step": 3492
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3518879302347156e-05,
-      "loss": 0.0314,
+      "loss": 0.2196,
       "step": 3493
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.351702384265702e-05,
-      "loss": 0.1384,
+      "loss": 0.0895,
       "step": 3494
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3515168382966882e-05,
-      "loss": 0.3039,
+      "loss": 0.0182,
       "step": 3495
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3513312923276742e-05,
-      "loss": 0.0874,
+      "loss": 0.0545,
       "step": 3496
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3511457463586604e-05,
-      "loss": 0.1403,
+      "loss": 0.0946,
       "step": 3497
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3509602003896467e-05,
-      "loss": 0.0877,
+      "loss": 0.0706,
       "step": 3498
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3507746544206328e-05,
-      "loss": 0.0849,
+      "loss": 0.0577,
       "step": 3499
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.350589108451619e-05,
-      "loss": 0.1372,
+      "loss": 0.1129,
       "step": 3500
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3504035624826051e-05,
-      "loss": 0.1978,
+      "loss": 0.0593,
       "step": 3501
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3502180165135915e-05,
-      "loss": 0.245,
+      "loss": 0.2103,
       "step": 3502
     },
     {
       "epoch": 0.97,
       "learning_rate": 1.3500324705445775e-05,
-      "loss": 0.1931,
+      "loss": 0.0643,
       "step": 3503
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3498469245755637e-05,
-      "loss": 0.3005,
+      "loss": 0.0627,
       "step": 3504
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3496613786065497e-05,
-      "loss": 0.0313,
+      "loss": 0.0613,
       "step": 3505
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.349475832637536e-05,
-      "loss": 0.1399,
+      "loss": 0.1242,
       "step": 3506
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3492902866685223e-05,
-      "loss": 0.0879,
+      "loss": 0.099,
       "step": 3507
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3491047406995084e-05,
-      "loss": 0.1889,
+      "loss": 0.0185,
       "step": 3508
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3489191947304945e-05,
-      "loss": 0.1378,
+      "loss": 0.0192,
       "step": 3509
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3487336487614808e-05,
-      "loss": 0.0906,
+      "loss": 0.1411,
       "step": 3510
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.348548102792467e-05,
-      "loss": 0.0892,
+      "loss": 0.0196,
       "step": 3511
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.348362556823453e-05,
-      "loss": 0.143,
+      "loss": 0.095,
       "step": 3512
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3481770108544392e-05,
-      "loss": 0.0884,
+      "loss": 0.0725,
       "step": 3513
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3479914648854256e-05,
-      "loss": 0.1385,
+      "loss": 0.016,
       "step": 3514
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3478059189164116e-05,
-      "loss": 0.1411,
+      "loss": 0.1284,
       "step": 3515
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3476203729473978e-05,
-      "loss": 0.0868,
+      "loss": 0.1985,
       "step": 3516
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.347434826978384e-05,
-      "loss": 0.0859,
+      "loss": 0.089,
       "step": 3517
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3472492810093703e-05,
-      "loss": 0.0324,
+      "loss": 0.0202,
       "step": 3518
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3470637350403563e-05,
-      "loss": 0.1945,
+      "loss": 0.0543,
       "step": 3519
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3468781890713425e-05,
-      "loss": 0.0883,
+      "loss": 0.0529,
       "step": 3520
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3466926431023285e-05,
-      "loss": 0.1401,
+      "loss": 0.0326,
       "step": 3521
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3465070971333149e-05,
-      "loss": 0.1959,
+      "loss": 0.0555,
       "step": 3522
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.346321551164301e-05,
-      "loss": 0.0297,
+      "loss": 0.0918,
       "step": 3523
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3461360051952873e-05,
-      "loss": 0.1369,
+      "loss": 0.056,
       "step": 3524
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3459504592262733e-05,
-      "loss": 0.1929,
+      "loss": 0.0262,
       "step": 3525
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3457649132572596e-05,
-      "loss": 0.1372,
+      "loss": 0.145,
       "step": 3526
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3455793672882458e-05,
-      "loss": 0.0843,
+      "loss": 0.2898,
       "step": 3527
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3453938213192318e-05,
-      "loss": 0.1398,
+      "loss": 0.0646,
       "step": 3528
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.345208275350218e-05,
-      "loss": 0.0275,
+      "loss": 0.0091,
       "step": 3529
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3450227293812044e-05,
-      "loss": 0.0844,
+      "loss": 0.0535,
       "step": 3530
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3448371834121906e-05,
-      "loss": 0.0838,
+      "loss": 0.0679,
       "step": 3531
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3446516374431766e-05,
-      "loss": 0.1427,
+      "loss": 0.0908,
       "step": 3532
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3444660914741628e-05,
-      "loss": 0.3025,
+      "loss": 0.0977,
       "step": 3533
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3442805455051491e-05,
-      "loss": 0.028,
+      "loss": 0.0465,
       "step": 3534
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3440949995361352e-05,
-      "loss": 0.0824,
+      "loss": 0.1136,
       "step": 3535
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3439094535671213e-05,
-      "loss": 0.2548,
+      "loss": 0.101,
       "step": 3536
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3437239075981074e-05,
-      "loss": 0.252,
+      "loss": 0.0954,
       "step": 3537
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3435383616290937e-05,
-      "loss": 0.0836,
+      "loss": 0.0843,
       "step": 3538
     },
     {
       "epoch": 0.98,
       "learning_rate": 1.3433528156600799e-05,
-      "loss": 0.0271,
+      "loss": 0.0544,
       "step": 3539
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3431672696910661e-05,
-      "loss": 0.1385,
+      "loss": 0.0256,
       "step": 3540
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3429817237220521e-05,
-      "loss": 0.0278,
+      "loss": 0.06,
       "step": 3541
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3427961777530385e-05,
-      "loss": 0.0854,
+      "loss": 0.02,
       "step": 3542
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3426106317840247e-05,
-      "loss": 0.0852,
+      "loss": 0.1058,
       "step": 3543
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3424250858150107e-05,
-      "loss": 0.0833,
+      "loss": 0.2259,
       "step": 3544
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3422395398459969e-05,
-      "loss": 0.0276,
+      "loss": 0.0209,
       "step": 3545
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3420539938769832e-05,
-      "loss": 0.0854,
+      "loss": 0.0153,
       "step": 3546
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3418684479079694e-05,
-      "loss": 0.0838,
+      "loss": 0.0203,
       "step": 3547
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3416829019389554e-05,
-      "loss": 0.0832,
+      "loss": 0.0659,
       "step": 3548
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3414973559699416e-05,
-      "loss": 0.1411,
+      "loss": 0.2351,
       "step": 3549
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.341311810000928e-05,
-      "loss": 0.083,
+      "loss": 0.0955,
       "step": 3550
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.341126264031914e-05,
-      "loss": 0.1972,
+      "loss": 0.068,
       "step": 3551
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3409407180629002e-05,
-      "loss": 0.0254,
+      "loss": 0.1686,
       "step": 3552
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3407551720938864e-05,
-      "loss": 0.0231,
+      "loss": 0.0088,
       "step": 3553
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3405696261248725e-05,
-      "loss": 0.0808,
+      "loss": 0.065,
       "step": 3554
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3403840801558587e-05,
-      "loss": 0.2561,
+      "loss": 0.0766,
       "step": 3555
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.340198534186845e-05,
-      "loss": 0.1996,
+      "loss": 0.0211,
       "step": 3556
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.340012988217831e-05,
-      "loss": 0.0821,
+      "loss": 0.0241,
       "step": 3557
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3398274422488173e-05,
-      "loss": 0.0811,
+      "loss": 0.0776,
       "step": 3558
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3396418962798035e-05,
-      "loss": 0.1391,
+      "loss": 0.016,
       "step": 3559
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3394563503107895e-05,
-      "loss": 0.1929,
+      "loss": 0.0786,
       "step": 3560
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3392708043417757e-05,
-      "loss": 0.2025,
+      "loss": 0.1187,
       "step": 3561
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.339085258372762e-05,
-      "loss": 0.1393,
+      "loss": 0.2772,
       "step": 3562
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3388997124037482e-05,
-      "loss": 0.0849,
+      "loss": 0.0181,
       "step": 3563
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3387141664347342e-05,
-      "loss": 0.0232,
+      "loss": 0.0759,
       "step": 3564
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3385286204657204e-05,
-      "loss": 0.1371,
+      "loss": 0.0224,
       "step": 3565
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3383430744967068e-05,
-      "loss": 0.1982,
+      "loss": 0.0587,
       "step": 3566
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3381575285276928e-05,
-      "loss": 0.2008,
+      "loss": 0.0149,
       "step": 3567
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.337971982558679e-05,
-      "loss": 0.0822,
+      "loss": 0.0123,
       "step": 3568
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3377864365896652e-05,
-      "loss": 0.1973,
+      "loss": 0.1851,
       "step": 3569
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3376008906206515e-05,
-      "loss": 0.0255,
+      "loss": 0.0362,
       "step": 3570
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3374153446516376e-05,
-      "loss": 0.1394,
+      "loss": 0.0687,
       "step": 3571
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3372297986826237e-05,
-      "loss": 0.0808,
+      "loss": 0.1331,
       "step": 3572
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3370442527136098e-05,
-      "loss": 0.1394,
+      "loss": 0.2052,
       "step": 3573
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3368587067445961e-05,
-      "loss": 0.2551,
+      "loss": 0.1056,
       "step": 3574
     },
     {
       "epoch": 0.99,
       "learning_rate": 1.3366731607755823e-05,
-      "loss": 0.1438,
+      "loss": 0.02,
       "step": 3575
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3364876148065685e-05,
-      "loss": 0.1408,
+      "loss": 0.1132,
       "step": 3576
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3363020688375545e-05,
-      "loss": 0.142,
+      "loss": 0.0714,
       "step": 3577
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3361165228685409e-05,
-      "loss": 0.1431,
+      "loss": 0.0885,
       "step": 3578
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.335930976899527e-05,
-      "loss": 0.3031,
+      "loss": 0.0602,
       "step": 3579
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.335745430930513e-05,
-      "loss": 0.03,
+      "loss": 0.1419,
       "step": 3580
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3355598849614993e-05,
-      "loss": 0.0291,
+      "loss": 0.1483,
       "step": 3581
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3353743389924856e-05,
-      "loss": 0.1958,
+      "loss": 0.2047,
       "step": 3582
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3351887930234716e-05,
-      "loss": 0.3051,
+      "loss": 0.1558,
       "step": 3583
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3350032470544578e-05,
-      "loss": 0.3044,
+      "loss": 0.0981,
       "step": 3584
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.334817701085444e-05,
-      "loss": 0.1934,
+      "loss": 0.0751,
       "step": 3585
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3346321551164304e-05,
-      "loss": 0.1385,
+      "loss": 0.1334,
       "step": 3586
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3344466091474164e-05,
-      "loss": 0.0344,
+      "loss": 0.1741,
       "step": 3587
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3342610631784026e-05,
-      "loss": 0.0841,
+      "loss": 0.0741,
       "step": 3588
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3340755172093886e-05,
-      "loss": 0.138,
+      "loss": 0.118,
       "step": 3589
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.333889971240375e-05,
-      "loss": 0.1417,
+      "loss": 0.0697,
       "step": 3590
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3337044252713611e-05,
-      "loss": 0.0346,
+      "loss": 0.0506,
       "step": 3591
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3335188793023473e-05,
-      "loss": 0.1959,
+      "loss": 0.0991,
       "step": 3592
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3333333333333333e-05,
-      "loss": 0.4093,
+      "loss": 0.0759,
       "step": 3593
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3331477873643197e-05,
-      "loss": 0.0882,
+      "loss": 0.067,
       "step": 3594
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3329622413953059e-05,
-      "loss": 0.1888,
+      "loss": 0.1084,
       "step": 3595
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3327766954262919e-05,
-      "loss": 0.1377,
+      "loss": 0.2516,
       "step": 3596
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.332591149457278e-05,
-      "loss": 0.139,
+      "loss": 0.0861,
       "step": 3597
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3324056034882644e-05,
-      "loss": 0.1403,
+      "loss": 0.0609,
       "step": 3598
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3322200575192506e-05,
-      "loss": 0.1379,
+      "loss": 0.081,
       "step": 3599
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3320345115502366e-05,
-      "loss": 0.1945,
+      "loss": 0.0417,
       "step": 3600
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3318489655812228e-05,
-      "loss": 0.1378,
+      "loss": 0.1091,
       "step": 3601
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3316634196122092e-05,
-      "loss": 0.2417,
+      "loss": 0.1901,
       "step": 3602
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3314778736431952e-05,
-      "loss": 0.1891,
+      "loss": 0.0713,
       "step": 3603
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3312923276741814e-05,
-      "loss": 0.0881,
+      "loss": 0.0247,
       "step": 3604
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3311067817051676e-05,
-      "loss": 0.1424,
+      "loss": 0.1091,
       "step": 3605
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3309212357361538e-05,
-      "loss": 0.1926,
+      "loss": 0.0189,
       "step": 3606
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.33073568976714e-05,
-      "loss": 0.0876,
+      "loss": 0.0256,
       "step": 3607
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3305501437981261e-05,
-      "loss": 0.1863,
+      "loss": 0.0873,
       "step": 3608
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3303645978291122e-05,
-      "loss": 0.1414,
+      "loss": 0.0188,
       "step": 3609
     },
     {
       "epoch": 1.0,
       "learning_rate": 1.3301790518600985e-05,
-      "loss": 0.2941,
+      "loss": 0.0248,
       "step": 3610
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3299935058910847e-05,
-      "loss": 0.1912,
+      "loss": 0.0845,
       "step": 3611
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3298079599220707e-05,
-      "loss": 0.087,
+      "loss": 0.0101,
       "step": 3612
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3296224139530569e-05,
-      "loss": 0.338,
+      "loss": 0.0515,
       "step": 3613
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3294368679840433e-05,
-      "loss": 0.1375,
+      "loss": 0.0145,
       "step": 3614
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3292513220150294e-05,
-      "loss": 0.1968,
+      "loss": 0.0081,
       "step": 3615
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3290657760460155e-05,
-      "loss": 0.286,
+      "loss": 0.1699,
       "step": 3616
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3288802300770016e-05,
-      "loss": 0.1382,
+      "loss": 0.0559,
       "step": 3617
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.328694684107988e-05,
-      "loss": 0.0932,
+      "loss": 0.0071,
       "step": 3618
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.328509138138974e-05,
-      "loss": 0.1865,
+      "loss": 0.0266,
       "step": 3619
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3283235921699602e-05,
-      "loss": 0.2364,
+      "loss": 0.0082,
       "step": 3620
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3281380462009464e-05,
-      "loss": 0.1436,
+      "loss": 0.0885,
       "step": 3621
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3279525002319328e-05,
-      "loss": 0.1473,
+      "loss": 0.0515,
       "step": 3622
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3277669542629188e-05,
-      "loss": 0.1387,
+      "loss": 0.0054,
       "step": 3623
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.327581408293905e-05,
-      "loss": 0.1944,
+      "loss": 0.0701,
       "step": 3624
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.327395862324891e-05,
-      "loss": 0.2368,
+      "loss": 0.0097,
       "step": 3625
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3272103163558773e-05,
-      "loss": 0.0944,
+      "loss": 0.1414,
       "step": 3626
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3270247703868635e-05,
-      "loss": 0.0457,
+      "loss": 0.0747,
       "step": 3627
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3268392244178497e-05,
-      "loss": 0.1888,
+      "loss": 0.0271,
       "step": 3628
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3266536784488357e-05,
-      "loss": 0.0926,
+      "loss": 0.009,
       "step": 3629
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.326468132479822e-05,
-      "loss": 0.1931,
+      "loss": 0.1751,
       "step": 3630
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3262825865108083e-05,
-      "loss": 0.0923,
+      "loss": 0.0684,
       "step": 3631
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3260970405417943e-05,
-      "loss": 0.1409,
+      "loss": 0.0956,
       "step": 3632
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3259114945727805e-05,
-      "loss": 0.137,
+      "loss": 0.0547,
       "step": 3633
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3257259486037668e-05,
-      "loss": 0.1401,
+      "loss": 0.0582,
       "step": 3634
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3255404026347528e-05,
-      "loss": 0.2927,
+      "loss": 0.0185,
       "step": 3635
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.325354856665739e-05,
-      "loss": 0.142,
+      "loss": 0.1134,
       "step": 3636
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3251693106967252e-05,
-      "loss": 0.2451,
+      "loss": 0.0372,
       "step": 3637
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3249837647277116e-05,
-      "loss": 0.0379,
+      "loss": 0.1949,
       "step": 3638
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3247982187586976e-05,
-      "loss": 0.1413,
+      "loss": 0.0139,
       "step": 3639
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3246126727896838e-05,
-      "loss": 0.0862,
+      "loss": 0.0219,
       "step": 3640
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3244271268206698e-05,
-      "loss": 0.0356,
+      "loss": 0.013,
       "step": 3641
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3242415808516562e-05,
-      "loss": 0.0902,
+      "loss": 0.0649,
       "step": 3642
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3240560348826423e-05,
-      "loss": 0.0867,
+      "loss": 0.0662,
       "step": 3643
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3238704889136285e-05,
-      "loss": 0.087,
+      "loss": 0.0633,
       "step": 3644
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3236849429446145e-05,
-      "loss": 0.1382,
+      "loss": 0.0204,
       "step": 3645
     },
     {
       "epoch": 1.01,
       "learning_rate": 1.3234993969756007e-05,
-      "loss": 0.0324,
+      "loss": 0.0629,
       "step": 3646
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3233138510065871e-05,
-      "loss": 0.1376,
+      "loss": 0.0517,
       "step": 3647
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3231283050375731e-05,
-      "loss": 0.2425,
+      "loss": 0.1309,
       "step": 3648
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3229427590685593e-05,
-      "loss": 0.3618,
+      "loss": 0.1694,
       "step": 3649
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3227572130995455e-05,
-      "loss": 0.1945,
+      "loss": 0.1513,
       "step": 3650
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3225716671305318e-05,
-      "loss": 0.1372,
+      "loss": 0.0142,
       "step": 3651
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3223861211615179e-05,
-      "loss": 0.0844,
+      "loss": 0.0764,
       "step": 3652
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.322200575192504e-05,
-      "loss": 0.2465,
+      "loss": 0.0926,
       "step": 3653
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.32201502922349e-05,
-      "loss": 0.1407,
+      "loss": 0.0616,
       "step": 3654
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3218294832544764e-05,
-      "loss": 0.0844,
+      "loss": 0.1134,
       "step": 3655
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3216439372854626e-05,
-      "loss": 0.0854,
+      "loss": 0.101,
       "step": 3656
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3214583913164486e-05,
-      "loss": 0.1404,
+      "loss": 0.0786,
       "step": 3657
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3212728453474348e-05,
-      "loss": 0.0314,
+      "loss": 0.0249,
       "step": 3658
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3210872993784212e-05,
-      "loss": 0.0318,
+      "loss": 0.1624,
       "step": 3659
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3209017534094074e-05,
-      "loss": 0.1387,
+      "loss": 0.0243,
       "step": 3660
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3207162074403934e-05,
-      "loss": 0.0302,
+      "loss": 0.0258,
       "step": 3661
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3205306614713796e-05,
-      "loss": 0.0868,
+      "loss": 0.0619,
       "step": 3662
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3203451155023659e-05,
-      "loss": 0.138,
+      "loss": 0.066,
       "step": 3663
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.320159569533352e-05,
-      "loss": 0.0841,
+      "loss": 0.0245,
       "step": 3664
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3199740235643381e-05,
-      "loss": 0.0836,
+      "loss": 0.1215,
       "step": 3665
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3197884775953243e-05,
-      "loss": 0.1383,
+      "loss": 0.1076,
       "step": 3666
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3196029316263107e-05,
-      "loss": 0.1424,
+      "loss": 0.1283,
       "step": 3667
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3194173856572967e-05,
-      "loss": 0.085,
+      "loss": 0.1034,
       "step": 3668
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3192318396882829e-05,
-      "loss": 0.0278,
+      "loss": 0.0287,
       "step": 3669
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3190462937192689e-05,
-      "loss": 0.2026,
+      "loss": 0.0227,
       "step": 3670
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3188607477502552e-05,
-      "loss": 0.1393,
+      "loss": 0.1412,
       "step": 3671
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3186752017812414e-05,
-      "loss": 0.1441,
+      "loss": 0.0683,
       "step": 3672
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3184896558122276e-05,
-      "loss": 0.0839,
+      "loss": 0.0699,
       "step": 3673
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3183041098432136e-05,
-      "loss": 0.1967,
+      "loss": 0.0933,
       "step": 3674
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3181185638742e-05,
-      "loss": 0.1411,
+      "loss": 0.0655,
       "step": 3675
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3179330179051862e-05,
-      "loss": 0.1394,
+      "loss": 0.2068,
       "step": 3676
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3177474719361722e-05,
-      "loss": 0.0825,
+      "loss": 0.0603,
       "step": 3677
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3175619259671584e-05,
-      "loss": 0.1355,
+      "loss": 0.0647,
       "step": 3678
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3173763799981447e-05,
-      "loss": 0.2003,
+      "loss": 0.1638,
       "step": 3679
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3171908340291308e-05,
-      "loss": 0.1989,
+      "loss": 0.269,
       "step": 3680
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.317005288060117e-05,
-      "loss": 0.0818,
+      "loss": 0.2056,
       "step": 3681
     },
     {
       "epoch": 1.02,
       "learning_rate": 1.3168197420911031e-05,
-      "loss": 0.1412,
+      "loss": 0.0781,
       "step": 3682
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3166341961220895e-05,
-      "loss": 0.1375,
+      "loss": 0.0182,
       "step": 3683
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3164486501530755e-05,
-      "loss": 0.1369,
+      "loss": 0.0115,
       "step": 3684
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3162631041840617e-05,
-      "loss": 0.1415,
+      "loss": 0.0218,
       "step": 3685
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3160775582150477e-05,
-      "loss": 0.1996,
+      "loss": 0.0292,
       "step": 3686
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.315892012246034e-05,
-      "loss": 0.2017,
+      "loss": 0.114,
       "step": 3687
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3157064662770203e-05,
-      "loss": 0.139,
+      "loss": 0.1259,
       "step": 3688
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3155209203080064e-05,
-      "loss": 0.1977,
+      "loss": 0.0618,
       "step": 3689
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3153353743389925e-05,
-      "loss": 0.1405,
+      "loss": 0.1108,
       "step": 3690
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3151498283699788e-05,
-      "loss": 0.4702,
+      "loss": 0.0717,
       "step": 3691
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.314964282400965e-05,
-      "loss": 0.1362,
+      "loss": 0.0512,
       "step": 3692
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.314778736431951e-05,
-      "loss": 0.0844,
+      "loss": 0.012,
       "step": 3693
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3145931904629372e-05,
-      "loss": 0.0866,
+      "loss": 0.0528,
       "step": 3694
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3144076444939236e-05,
-      "loss": 0.2499,
+      "loss": 0.0646,
       "step": 3695
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3142220985249097e-05,
-      "loss": 0.2508,
+      "loss": 0.1199,
       "step": 3696
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3140365525558958e-05,
-      "loss": 0.0849,
+      "loss": 0.0225,
       "step": 3697
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.313851006586882e-05,
-      "loss": 0.1407,
+      "loss": 0.09,
       "step": 3698
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3136654606178683e-05,
-      "loss": 0.0343,
+      "loss": 0.0697,
       "step": 3699
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3134799146488543e-05,
-      "loss": 0.1962,
+      "loss": 0.1004,
       "step": 3700
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3132943686798405e-05,
-      "loss": 0.0874,
+      "loss": 0.1302,
       "step": 3701
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3131088227108267e-05,
-      "loss": 0.2486,
+      "loss": 0.1097,
       "step": 3702
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3129232767418129e-05,
-      "loss": 0.0875,
+      "loss": 0.091,
       "step": 3703
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.312737730772799e-05,
-      "loss": 0.1409,
+      "loss": 0.0657,
       "step": 3704
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3125521848037853e-05,
-      "loss": 0.0353,
+      "loss": 0.0354,
       "step": 3705
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3123666388347713e-05,
-      "loss": 0.1932,
+      "loss": 0.0576,
       "step": 3706
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3121810928657576e-05,
-      "loss": 0.0347,
+      "loss": 0.1045,
       "step": 3707
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3119955468967438e-05,
-      "loss": 0.2439,
+      "loss": 0.0544,
       "step": 3708
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3118100009277298e-05,
-      "loss": 0.1895,
+      "loss": 0.0608,
       "step": 3709
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.311624454958716e-05,
-      "loss": 0.0898,
+      "loss": 0.0514,
       "step": 3710
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3114389089897024e-05,
-      "loss": 0.0866,
+      "loss": 0.0319,
       "step": 3711
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3112533630206886e-05,
-      "loss": 0.0863,
+      "loss": 0.1432,
       "step": 3712
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3110678170516746e-05,
-      "loss": 0.0881,
+      "loss": 0.0343,
       "step": 3713
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3108822710826608e-05,
-      "loss": 0.1434,
+      "loss": 0.105,
       "step": 3714
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3106967251136471e-05,
-      "loss": 0.1355,
+      "loss": 0.1345,
       "step": 3715
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3105111791446332e-05,
-      "loss": 0.2442,
+      "loss": 0.097,
       "step": 3716
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3103256331756193e-05,
-      "loss": 0.0334,
+      "loss": 0.0993,
       "step": 3717
     },
     {
       "epoch": 1.03,
       "learning_rate": 1.3101400872066055e-05,
-      "loss": 0.1871,
+      "loss": 0.0615,
       "step": 3718
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3099545412375919e-05,
-      "loss": 0.0332,
+      "loss": 0.0282,
       "step": 3719
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3097689952685779e-05,
-      "loss": 0.1971,
+      "loss": 0.1234,
       "step": 3720
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3095834492995641e-05,
-      "loss": 0.0882,
+      "loss": 0.0224,
       "step": 3721
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3093979033305501e-05,
-      "loss": 0.1903,
+      "loss": 0.1519,
       "step": 3722
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3092123573615365e-05,
-      "loss": 0.0866,
+      "loss": 0.053,
       "step": 3723
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3090268113925226e-05,
-      "loss": 0.1929,
+      "loss": 0.0185,
       "step": 3724
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3088412654235088e-05,
-      "loss": 0.1914,
+      "loss": 0.161,
       "step": 3725
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3086557194544949e-05,
-      "loss": 0.0305,
+      "loss": 0.0411,
       "step": 3726
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3084701734854812e-05,
-      "loss": 0.0317,
+      "loss": 0.096,
       "step": 3727
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3082846275164674e-05,
-      "loss": 0.0302,
+      "loss": 0.0529,
       "step": 3728
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3080990815474534e-05,
-      "loss": 0.1922,
+      "loss": 0.1079,
       "step": 3729
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3079135355784396e-05,
-      "loss": 0.1951,
+      "loss": 0.0938,
       "step": 3730
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.307727989609426e-05,
-      "loss": 0.0852,
+      "loss": 0.0805,
       "step": 3731
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.307542443640412e-05,
-      "loss": 0.1393,
+      "loss": 0.0606,
       "step": 3732
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3073568976713982e-05,
-      "loss": 0.1371,
+      "loss": 0.054,
       "step": 3733
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3071713517023843e-05,
-      "loss": 0.1423,
+      "loss": 0.0114,
       "step": 3734
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3069858057333707e-05,
-      "loss": 0.2001,
+      "loss": 0.04,
       "step": 3735
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3068002597643567e-05,
-      "loss": 0.1948,
+      "loss": 0.0166,
       "step": 3736
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3066147137953429e-05,
-      "loss": 0.3078,
+      "loss": 0.0551,
       "step": 3737
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.306429167826329e-05,
-      "loss": 0.1389,
+      "loss": 0.0459,
       "step": 3738
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3062436218573153e-05,
-      "loss": 0.0864,
+      "loss": 0.0773,
       "step": 3739
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3060580758883015e-05,
-      "loss": 0.1913,
+      "loss": 0.0751,
       "step": 3740
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3058725299192877e-05,
-      "loss": 0.2471,
+      "loss": 0.0889,
       "step": 3741
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3056869839502737e-05,
-      "loss": 0.084,
+      "loss": 0.1411,
       "step": 3742
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.30550143798126e-05,
-      "loss": 0.1392,
+      "loss": 0.2186,
       "step": 3743
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3053158920122462e-05,
-      "loss": 0.1924,
+      "loss": 0.0105,
       "step": 3744
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3051303460432322e-05,
-      "loss": 0.1385,
+      "loss": 0.1006,
       "step": 3745
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3049448000742184e-05,
-      "loss": 0.1411,
+      "loss": 0.0578,
       "step": 3746
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3047592541052048e-05,
-      "loss": 0.1901,
+      "loss": 0.0136,
       "step": 3747
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.304573708136191e-05,
-      "loss": 0.1362,
+      "loss": 0.2098,
       "step": 3748
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.304388162167177e-05,
-      "loss": 0.0906,
+      "loss": 0.06,
       "step": 3749
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3042026161981632e-05,
-      "loss": 0.0856,
+      "loss": 0.016,
       "step": 3750
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3040170702291495e-05,
-      "loss": 0.1924,
+      "loss": 0.1407,
       "step": 3751
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3038315242601355e-05,
-      "loss": 0.036,
+      "loss": 0.0189,
       "step": 3752
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.3036459782911217e-05,
-      "loss": 0.1427,
+      "loss": 0.0522,
       "step": 3753
     },
     {
       "epoch": 1.04,
       "learning_rate": 1.303460432322108e-05,
-      "loss": 0.189,
+      "loss": 0.114,
       "step": 3754
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3032748863530941e-05,
-      "loss": 0.0854,
+      "loss": 0.1436,
       "step": 3755
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3030893403840803e-05,
-      "loss": 0.1889,
+      "loss": 0.0472,
       "step": 3756
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3029037944150665e-05,
-      "loss": 0.0868,
+      "loss": 0.1825,
       "step": 3757
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3027182484460525e-05,
-      "loss": 0.1925,
+      "loss": 0.0849,
       "step": 3758
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3025327024770389e-05,
-      "loss": 0.1358,
+      "loss": 0.1088,
       "step": 3759
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.302347156508025e-05,
-      "loss": 0.1851,
+      "loss": 0.2055,
       "step": 3760
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.302161610539011e-05,
-      "loss": 0.0873,
+      "loss": 0.084,
       "step": 3761
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3019760645699972e-05,
-      "loss": 0.1403,
+      "loss": 0.0216,
       "step": 3762
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3017905186009836e-05,
-      "loss": 0.1941,
+      "loss": 0.0203,
       "step": 3763
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3016049726319698e-05,
-      "loss": 0.2375,
+      "loss": 0.1174,
       "step": 3764
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3014194266629558e-05,
-      "loss": 0.0344,
+      "loss": 0.0894,
       "step": 3765
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.301233880693942e-05,
-      "loss": 0.0836,
+      "loss": 0.0757,
       "step": 3766
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3010483347249284e-05,
-      "loss": 0.034,
+      "loss": 0.1171,
       "step": 3767
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3008627887559144e-05,
-      "loss": 0.0892,
+      "loss": 0.1634,
       "step": 3768
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3006772427869006e-05,
-      "loss": 0.1898,
+      "loss": 0.0235,
       "step": 3769
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3004916968178867e-05,
-      "loss": 0.3047,
+      "loss": 0.1299,
       "step": 3770
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3003061508488731e-05,
-      "loss": 0.0831,
+      "loss": 0.0181,
       "step": 3771
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.3001206048798591e-05,
-      "loss": 0.0313,
+      "loss": 0.0562,
       "step": 3772
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2999350589108453e-05,
-      "loss": 0.193,
+      "loss": 0.1046,
       "step": 3773
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2997495129418313e-05,
-      "loss": 0.0839,
+      "loss": 0.1089,
       "step": 3774
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2995639669728177e-05,
-      "loss": 0.0897,
+      "loss": 0.0553,
       "step": 3775
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2993784210038039e-05,
-      "loss": 0.1391,
+      "loss": 0.1611,
       "step": 3776
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2991928750347899e-05,
-      "loss": 0.2516,
+      "loss": 0.1451,
       "step": 3777
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.299007329065776e-05,
-      "loss": 0.2486,
+      "loss": 0.067,
       "step": 3778
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2988217830967624e-05,
-      "loss": 0.2989,
+      "loss": 0.0556,
       "step": 3779
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2986362371277486e-05,
-      "loss": 0.1425,
+      "loss": 0.0767,
       "step": 3780
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2984506911587346e-05,
-      "loss": 0.1949,
+      "loss": 0.0568,
       "step": 3781
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2982651451897208e-05,
-      "loss": 0.0849,
+      "loss": 0.0875,
       "step": 3782
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2980795992207072e-05,
-      "loss": 0.0338,
+      "loss": 0.1003,
       "step": 3783
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2978940532516932e-05,
-      "loss": 0.0853,
+      "loss": 0.103,
       "step": 3784
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2977085072826794e-05,
-      "loss": 0.2416,
+      "loss": 0.0164,
       "step": 3785
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2975229613136656e-05,
-      "loss": 0.1397,
+      "loss": 0.0201,
       "step": 3786
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.297337415344652e-05,
-      "loss": 0.0865,
+      "loss": 0.0574,
       "step": 3787
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.297151869375638e-05,
-      "loss": 0.0839,
+      "loss": 0.0339,
       "step": 3788
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2969663234066241e-05,
-      "loss": 0.0865,
+      "loss": 0.0651,
       "step": 3789
     },
     {
       "epoch": 1.05,
       "learning_rate": 1.2967807774376101e-05,
-      "loss": 0.1939,
+      "loss": 0.0531,
       "step": 3790
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2965952314685965e-05,
-      "loss": 0.0346,
+      "loss": 0.0142,
       "step": 3791
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2964096854995827e-05,
-      "loss": 0.1898,
+      "loss": 0.0485,
       "step": 3792
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2962241395305689e-05,
-      "loss": 0.0321,
+      "loss": 0.0365,
       "step": 3793
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2960385935615549e-05,
-      "loss": 0.1904,
+      "loss": 0.0561,
       "step": 3794
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2958530475925413e-05,
-      "loss": 0.0316,
+      "loss": 0.1146,
       "step": 3795
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2956675016235274e-05,
-      "loss": 0.0864,
+      "loss": 0.0811,
       "step": 3796
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2954819556545135e-05,
-      "loss": 0.198,
+      "loss": 0.133,
       "step": 3797
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2952964096854996e-05,
-      "loss": 0.2462,
+      "loss": 0.0561,
       "step": 3798
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2951108637164858e-05,
-      "loss": 0.2462,
+      "loss": 0.0816,
       "step": 3799
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.294925317747472e-05,
-      "loss": 0.0306,
+      "loss": 0.0925,
       "step": 3800
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2947397717784582e-05,
-      "loss": 0.1963,
+      "loss": 0.0444,
       "step": 3801
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2945542258094444e-05,
-      "loss": 0.0843,
+      "loss": 0.1219,
       "step": 3802
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2943686798404304e-05,
-      "loss": 0.2513,
+      "loss": 0.0999,
       "step": 3803
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2941831338714168e-05,
-      "loss": 0.3001,
+      "loss": 0.0748,
       "step": 3804
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.293997587902403e-05,
-      "loss": 0.0845,
+      "loss": 0.1197,
       "step": 3805
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.293812041933389e-05,
-      "loss": 0.1385,
+      "loss": 0.1355,
       "step": 3806
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2936264959643752e-05,
-      "loss": 0.084,
+      "loss": 0.0189,
       "step": 3807
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2934409499953615e-05,
-      "loss": 0.1924,
+      "loss": 0.0184,
       "step": 3808
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2932554040263477e-05,
-      "loss": 0.1361,
+      "loss": 0.0995,
       "step": 3809
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2930698580573337e-05,
-      "loss": 0.1313,
+      "loss": 0.03,
       "step": 3810
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2928843120883199e-05,
-      "loss": 0.0331,
+      "loss": 0.0185,
       "step": 3811
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2926987661193063e-05,
-      "loss": 0.2923,
+      "loss": 0.0942,
       "step": 3812
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2925132201502923e-05,
-      "loss": 0.1393,
+      "loss": 0.0465,
       "step": 3813
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2923276741812785e-05,
-      "loss": 0.0842,
+      "loss": 0.0104,
       "step": 3814
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2921421282122647e-05,
-      "loss": 0.0825,
+      "loss": 0.0112,
       "step": 3815
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.291956582243251e-05,
-      "loss": 0.1373,
+      "loss": 0.1109,
       "step": 3816
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.291771036274237e-05,
-      "loss": 0.1355,
+      "loss": 0.2626,
       "step": 3817
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2915854903052232e-05,
-      "loss": 0.1369,
+      "loss": 0.1373,
       "step": 3818
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2913999443362092e-05,
-      "loss": 0.0838,
+      "loss": 0.0119,
       "step": 3819
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2912143983671956e-05,
-      "loss": 0.1386,
+      "loss": 0.0131,
       "step": 3820
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2910288523981818e-05,
-      "loss": 0.0886,
+      "loss": 0.0971,
       "step": 3821
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.290843306429168e-05,
-      "loss": 0.1923,
+      "loss": 0.0613,
       "step": 3822
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.290657760460154e-05,
-      "loss": 0.2466,
+      "loss": 0.0566,
       "step": 3823
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2904722144911403e-05,
-      "loss": 0.1919,
+      "loss": 0.0102,
       "step": 3824
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2902866685221265e-05,
-      "loss": 0.1957,
+      "loss": 0.0096,
       "step": 3825
     },
     {
       "epoch": 1.06,
       "learning_rate": 1.2901011225531125e-05,
-      "loss": 0.1352,
+      "loss": 0.0884,
       "step": 3826
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2899155765840987e-05,
-      "loss": 0.1368,
+      "loss": 0.0092,
       "step": 3827
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2897300306150851e-05,
-      "loss": 0.1401,
+      "loss": 0.011,
       "step": 3828
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2895444846460711e-05,
-      "loss": 0.1863,
+      "loss": 0.0066,
       "step": 3829
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2893589386770573e-05,
-      "loss": 0.0852,
+      "loss": 0.0089,
       "step": 3830
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2891733927080435e-05,
-      "loss": 0.1428,
+      "loss": 0.0595,
       "step": 3831
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2889878467390298e-05,
-      "loss": 0.1905,
+      "loss": 0.0827,
       "step": 3832
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2888023007700159e-05,
-      "loss": 0.1447,
+      "loss": 0.1333,
       "step": 3833
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.288616754801002e-05,
-      "loss": 0.1863,
+      "loss": 0.0754,
       "step": 3834
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.288431208831988e-05,
-      "loss": 0.0903,
+      "loss": 0.0185,
       "step": 3835
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2882456628629744e-05,
-      "loss": 0.0855,
+      "loss": 0.0915,
       "step": 3836
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2880601168939606e-05,
-      "loss": 0.1402,
+      "loss": 0.0064,
       "step": 3837
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2878745709249468e-05,
-      "loss": 0.0882,
+      "loss": 0.0859,
       "step": 3838
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2876890249559328e-05,
-      "loss": 0.0344,
+      "loss": 0.1047,
       "step": 3839
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2875034789869192e-05,
-      "loss": 0.3016,
+      "loss": 0.1557,
       "step": 3840
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2873179330179053e-05,
-      "loss": 0.137,
+      "loss": 0.1363,
       "step": 3841
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2871323870488914e-05,
-      "loss": 0.1448,
+      "loss": 0.0134,
       "step": 3842
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2869468410798776e-05,
-      "loss": 0.0839,
+      "loss": 0.1839,
       "step": 3843
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2867612951108639e-05,
-      "loss": 0.0862,
+      "loss": 0.0925,
       "step": 3844
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2865757491418501e-05,
-      "loss": 0.0317,
+      "loss": 0.1369,
       "step": 3845
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2863902031728361e-05,
-      "loss": 0.1404,
+      "loss": 0.0362,
       "step": 3846
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2862046572038223e-05,
-      "loss": 0.1423,
+      "loss": 0.0238,
       "step": 3847
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2860191112348087e-05,
-      "loss": 0.0838,
+      "loss": 0.1847,
       "step": 3848
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2858335652657947e-05,
-      "loss": 0.1373,
+      "loss": 0.0134,
       "step": 3849
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2856480192967809e-05,
-      "loss": 0.0301,
+      "loss": 0.0924,
       "step": 3850
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.285462473327767e-05,
-      "loss": 0.0869,
+      "loss": 0.1206,
       "step": 3851
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2852769273587532e-05,
-      "loss": 0.0864,
+      "loss": 0.0614,
       "step": 3852
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2850913813897394e-05,
-      "loss": 0.0844,
+      "loss": 0.0175,
       "step": 3853
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2849058354207256e-05,
-      "loss": 0.1978,
+      "loss": 0.0853,
       "step": 3854
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2847202894517116e-05,
-      "loss": 0.0269,
+      "loss": 0.027,
       "step": 3855
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.284534743482698e-05,
-      "loss": 0.0268,
+      "loss": 0.0247,
       "step": 3856
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2843491975136842e-05,
-      "loss": 0.2585,
+      "loss": 0.0509,
       "step": 3857
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2841636515446702e-05,
-      "loss": 0.0839,
+      "loss": 0.1695,
       "step": 3858
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2839781055756564e-05,
-      "loss": 0.1398,
+      "loss": 0.0277,
       "step": 3859
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.2837925596066427e-05,
-      "loss": 0.3029,
+      "loss": 0.0169,
       "step": 3860
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.283607013637629e-05,
-      "loss": 0.0827,
+      "loss": 0.2458,
       "step": 3861
     },
     {
       "epoch": 1.07,
       "learning_rate": 1.283421467668615e-05,
-      "loss": 0.0841,
+      "loss": 0.0186,
       "step": 3862
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2832359216996011e-05,
-      "loss": 0.2509,
+      "loss": 0.0164,
       "step": 3863
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2830503757305875e-05,
-      "loss": 0.0833,
+      "loss": 0.0108,
       "step": 3864
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2828648297615735e-05,
-      "loss": 0.0837,
+      "loss": 0.1172,
       "step": 3865
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2826792837925597e-05,
-      "loss": 0.1981,
+      "loss": 0.1065,
       "step": 3866
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2824937378235459e-05,
-      "loss": 0.14,
+      "loss": 0.1022,
       "step": 3867
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2823081918545322e-05,
-      "loss": 0.0838,
+      "loss": 0.0145,
       "step": 3868
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2821226458855182e-05,
-      "loss": 0.1984,
+      "loss": 0.1176,
       "step": 3869
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2819370999165044e-05,
-      "loss": 0.1961,
+      "loss": 0.0736,
       "step": 3870
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2817515539474905e-05,
-      "loss": 0.0263,
+      "loss": 0.1379,
       "step": 3871
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2815660079784768e-05,
-      "loss": 0.0265,
+      "loss": 0.0335,
       "step": 3872
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.281380462009463e-05,
-      "loss": 0.0268,
+      "loss": 0.0763,
       "step": 3873
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2811949160404492e-05,
-      "loss": 0.0842,
+      "loss": 0.082,
       "step": 3874
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2810093700714352e-05,
-      "loss": 0.0823,
+      "loss": 0.0175,
       "step": 3875
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2808238241024216e-05,
-      "loss": 0.1388,
+      "loss": 0.0625,
       "step": 3876
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2806382781334077e-05,
-      "loss": 0.1928,
+      "loss": 0.1443,
       "step": 3877
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2804527321643938e-05,
-      "loss": 0.0821,
+      "loss": 0.0483,
       "step": 3878
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.28026718619538e-05,
-      "loss": 0.1382,
+      "loss": 0.0174,
       "step": 3879
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2800816402263663e-05,
-      "loss": 0.1406,
+      "loss": 0.0218,
       "step": 3880
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2798960942573523e-05,
-      "loss": 0.0822,
+      "loss": 0.0116,
       "step": 3881
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2797105482883385e-05,
-      "loss": 0.0258,
+      "loss": 0.0469,
       "step": 3882
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2795250023193247e-05,
-      "loss": 0.1376,
+      "loss": 0.122,
       "step": 3883
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.279339456350311e-05,
-      "loss": 0.0253,
+      "loss": 0.1724,
       "step": 3884
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.279153910381297e-05,
-      "loss": 0.0817,
+      "loss": 0.0605,
       "step": 3885
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2789683644122833e-05,
-      "loss": 0.1385,
+      "loss": 0.1414,
       "step": 3886
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2787828184432693e-05,
-      "loss": 0.144,
+      "loss": 0.0248,
       "step": 3887
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2785972724742556e-05,
-      "loss": 0.1997,
+      "loss": 0.1055,
       "step": 3888
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2784117265052418e-05,
-      "loss": 0.0244,
+      "loss": 0.1515,
       "step": 3889
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.278226180536228e-05,
-      "loss": 0.0807,
+      "loss": 0.1934,
       "step": 3890
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.278040634567214e-05,
-      "loss": 0.2595,
+      "loss": 0.0493,
       "step": 3891
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2778550885982004e-05,
-      "loss": 0.024,
+      "loss": 0.0974,
       "step": 3892
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2776695426291866e-05,
-      "loss": 0.2547,
+      "loss": 0.0336,
       "step": 3893
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2774839966601726e-05,
-      "loss": 0.081,
+      "loss": 0.0602,
       "step": 3894
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2772984506911588e-05,
-      "loss": 0.0253,
+      "loss": 0.0649,
       "step": 3895
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2771129047221451e-05,
-      "loss": 0.2541,
+      "loss": 0.1293,
       "step": 3896
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2769273587531311e-05,
-      "loss": 0.0796,
+      "loss": 0.1628,
       "step": 3897
     },
     {
       "epoch": 1.08,
       "learning_rate": 1.2767418127841173e-05,
-      "loss": 0.2512,
+      "loss": 0.0193,
       "step": 3898
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2765562668151035e-05,
-      "loss": 0.1362,
+      "loss": 0.112,
       "step": 3899
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2763707208460899e-05,
-      "loss": 0.1398,
+      "loss": 0.0254,
       "step": 3900
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2761851748770759e-05,
-      "loss": 0.1397,
+      "loss": 0.0213,
       "step": 3901
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.275999628908062e-05,
-      "loss": 0.2573,
+      "loss": 0.0835,
       "step": 3902
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2758140829390481e-05,
-      "loss": 0.1966,
+      "loss": 0.0217,
       "step": 3903
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2756285369700345e-05,
-      "loss": 0.0814,
+      "loss": 0.1028,
       "step": 3904
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2754429910010206e-05,
-      "loss": 0.0824,
+      "loss": 0.1077,
       "step": 3905
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2752574450320068e-05,
-      "loss": 0.418,
+      "loss": 0.0157,
       "step": 3906
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2750718990629928e-05,
-      "loss": 0.2499,
+      "loss": 0.1502,
       "step": 3907
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2748863530939792e-05,
-      "loss": 0.1438,
+      "loss": 0.0275,
       "step": 3908
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2747008071249654e-05,
-      "loss": 0.0847,
+      "loss": 0.1275,
       "step": 3909
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2745152611559514e-05,
-      "loss": 0.0836,
+      "loss": 0.0795,
       "step": 3910
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2743297151869376e-05,
-      "loss": 0.1394,
+      "loss": 0.1065,
       "step": 3911
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.274144169217924e-05,
-      "loss": 0.087,
+      "loss": 0.0103,
       "step": 3912
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2739586232489101e-05,
-      "loss": 0.1369,
+      "loss": 0.09,
       "step": 3913
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2737730772798962e-05,
-      "loss": 0.1373,
+      "loss": 0.0876,
       "step": 3914
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2735875313108823e-05,
-      "loss": 0.0836,
+      "loss": 0.0139,
       "step": 3915
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2734019853418687e-05,
-      "loss": 0.1405,
+      "loss": 0.0517,
       "step": 3916
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2732164393728547e-05,
-      "loss": 0.032,
+      "loss": 0.012,
       "step": 3917
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2730308934038409e-05,
-      "loss": 0.0855,
+      "loss": 0.1016,
       "step": 3918
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2728453474348271e-05,
-      "loss": 0.0842,
+      "loss": 0.1538,
       "step": 3919
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2726598014658133e-05,
-      "loss": 0.0866,
+      "loss": 0.0757,
       "step": 3920
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2724742554967995e-05,
-      "loss": 0.1406,
+      "loss": 0.0484,
       "step": 3921
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2722887095277857e-05,
-      "loss": 0.0308,
+      "loss": 0.12,
       "step": 3922
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2721031635587717e-05,
-      "loss": 0.0861,
+      "loss": 0.1204,
       "step": 3923
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.271917617589758e-05,
-      "loss": 0.1423,
+      "loss": 0.056,
       "step": 3924
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2717320716207442e-05,
-      "loss": 0.083,
+      "loss": 0.1044,
       "step": 3925
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2715465256517302e-05,
-      "loss": 0.085,
+      "loss": 0.0527,
       "step": 3926
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2713609796827164e-05,
-      "loss": 0.2454,
+      "loss": 0.0567,
       "step": 3927
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2711754337137028e-05,
-      "loss": 0.2496,
+      "loss": 0.1662,
       "step": 3928
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.270989887744689e-05,
-      "loss": 0.1406,
+      "loss": 0.1695,
       "step": 3929
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.270804341775675e-05,
-      "loss": 0.028,
+      "loss": 0.1562,
       "step": 3930
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2706187958066612e-05,
-      "loss": 0.1449,
+      "loss": 0.0612,
       "step": 3931
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2704332498376475e-05,
-      "loss": 0.0855,
+      "loss": 0.1055,
       "step": 3932
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2702477038686335e-05,
-      "loss": 0.0279,
+      "loss": 0.0371,
       "step": 3933
     },
     {
       "epoch": 1.09,
       "learning_rate": 1.2700621578996197e-05,
-      "loss": 0.2471,
+      "loss": 0.0331,
       "step": 3934
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2698766119306059e-05,
-      "loss": 0.2498,
+      "loss": 0.1566,
       "step": 3935
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2696910659615923e-05,
-      "loss": 0.0825,
+      "loss": 0.13,
       "step": 3936
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2695055199925783e-05,
-      "loss": 0.0865,
+      "loss": 0.0281,
       "step": 3937
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2693199740235645e-05,
-      "loss": 0.1422,
+      "loss": 0.0566,
       "step": 3938
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2691344280545505e-05,
-      "loss": 0.2506,
+      "loss": 0.029,
       "step": 3939
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2689488820855369e-05,
-      "loss": 0.0865,
+      "loss": 0.024,
       "step": 3940
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.268763336116523e-05,
-      "loss": 0.0821,
+      "loss": 0.083,
       "step": 3941
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2685777901475092e-05,
-      "loss": 0.2468,
+      "loss": 0.0692,
       "step": 3942
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2683922441784952e-05,
-      "loss": 0.0838,
+      "loss": 0.0987,
       "step": 3943
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2682066982094816e-05,
-      "loss": 0.1434,
+      "loss": 0.0196,
       "step": 3944
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2680211522404678e-05,
-      "loss": 0.1375,
+      "loss": 0.0617,
       "step": 3945
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2678356062714538e-05,
-      "loss": 0.029,
+      "loss": 0.0905,
       "step": 3946
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.26765006030244e-05,
-      "loss": 0.1413,
+      "loss": 0.1167,
       "step": 3947
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2674645143334262e-05,
-      "loss": 0.1449,
+      "loss": 0.0661,
       "step": 3948
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2672789683644124e-05,
-      "loss": 0.0283,
+      "loss": 0.0247,
       "step": 3949
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2670934223953986e-05,
-      "loss": 0.2555,
+      "loss": 0.1133,
       "step": 3950
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2669078764263847e-05,
-      "loss": 0.0841,
+      "loss": 0.0835,
       "step": 3951
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2667223304573708e-05,
-      "loss": 0.0869,
+      "loss": 0.0121,
       "step": 3952
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2665367844883571e-05,
-      "loss": 0.0834,
+      "loss": 0.0513,
       "step": 3953
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2663512385193433e-05,
-      "loss": 0.2489,
+      "loss": 0.0969,
       "step": 3954
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2661656925503293e-05,
-      "loss": 0.1403,
+      "loss": 0.1016,
       "step": 3955
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2659801465813155e-05,
-      "loss": 0.2516,
+      "loss": 0.1051,
       "step": 3956
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2657946006123019e-05,
-      "loss": 0.1913,
+      "loss": 0.1122,
       "step": 3957
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.265609054643288e-05,
-      "loss": 0.2437,
+      "loss": 0.1417,
       "step": 3958
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.265423508674274e-05,
-      "loss": 0.1883,
+      "loss": 0.0591,
       "step": 3959
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2652379627052603e-05,
-      "loss": 0.1971,
+      "loss": 0.1789,
       "step": 3960
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2650524167362466e-05,
-      "loss": 0.0842,
+      "loss": 0.0173,
       "step": 3961
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2648668707672326e-05,
-      "loss": 0.0848,
+      "loss": 0.0188,
       "step": 3962
     },
     {
@@ -23788,40912 +23788,40912 @@
     {
       "epoch": 1.1,
       "learning_rate": 1.264495778829205e-05,
-      "loss": 0.0879,
+      "loss": 0.0825,
       "step": 3964
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2643102328601914e-05,
-      "loss": 0.084,
+      "loss": 0.1889,
       "step": 3965
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2641246868911774e-05,
-      "loss": 0.138,
+      "loss": 0.0322,
       "step": 3966
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2639391409221636e-05,
-      "loss": 0.0324,
+      "loss": 0.0707,
       "step": 3967
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2637535949531496e-05,
-      "loss": 0.0857,
+      "loss": 0.1363,
       "step": 3968
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.263568048984136e-05,
-      "loss": 0.0866,
+      "loss": 0.0275,
       "step": 3969
     },
     {
       "epoch": 1.1,
       "learning_rate": 1.2633825030151221e-05,
-      "loss": 0.1411,
+      "loss": 0.0217,
       "step": 3970
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2631969570461083e-05,
-      "loss": 0.1366,
+      "loss": 0.0351,
       "step": 3971
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2630114110770943e-05,
-      "loss": 0.0312,
+      "loss": 0.0951,
       "step": 3972
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2628258651080807e-05,
-      "loss": 0.1926,
+      "loss": 0.1691,
       "step": 3973
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2626403191390669e-05,
-      "loss": 0.0304,
+      "loss": 0.0732,
       "step": 3974
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2624547731700529e-05,
-      "loss": 0.0845,
+      "loss": 0.0552,
       "step": 3975
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.262269227201039e-05,
-      "loss": 0.1414,
+      "loss": 0.0224,
       "step": 3976
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2620836812320254e-05,
-      "loss": 0.0849,
+      "loss": 0.0606,
       "step": 3977
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2618981352630115e-05,
-      "loss": 0.0822,
+      "loss": 0.0609,
       "step": 3978
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2617125892939976e-05,
-      "loss": 0.082,
+      "loss": 0.0175,
       "step": 3979
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2615270433249838e-05,
-      "loss": 0.084,
+      "loss": 0.0988,
       "step": 3980
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2613414973559702e-05,
-      "loss": 0.2545,
+      "loss": 0.0756,
       "step": 3981
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2611559513869562e-05,
-      "loss": 0.1409,
+      "loss": 0.0891,
       "step": 3982
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2609704054179424e-05,
-      "loss": 0.0844,
+      "loss": 0.022,
       "step": 3983
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2607848594489284e-05,
-      "loss": 0.3597,
+      "loss": 0.0497,
       "step": 3984
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2605993134799148e-05,
-      "loss": 0.0848,
+      "loss": 0.0301,
       "step": 3985
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.260413767510901e-05,
-      "loss": 0.0805,
+      "loss": 0.0569,
       "step": 3986
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2602282215418871e-05,
-      "loss": 0.0843,
+      "loss": 0.0132,
       "step": 3987
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2600426755728732e-05,
-      "loss": 0.2019,
+      "loss": 0.1112,
       "step": 3988
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2598571296038595e-05,
-      "loss": 0.0816,
+      "loss": 0.0157,
       "step": 3989
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2596715836348457e-05,
-      "loss": 0.1409,
+      "loss": 0.1442,
       "step": 3990
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2594860376658317e-05,
-      "loss": 0.0809,
+      "loss": 0.0559,
       "step": 3991
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2593004916968179e-05,
-      "loss": 0.1944,
+      "loss": 0.0143,
       "step": 3992
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2591149457278043e-05,
-      "loss": 0.1948,
+      "loss": 0.0684,
       "step": 3993
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2589293997587904e-05,
-      "loss": 0.1401,
+      "loss": 0.0182,
       "step": 3994
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2587438537897765e-05,
-      "loss": 0.3086,
+      "loss": 0.1145,
       "step": 3995
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2585583078207626e-05,
-      "loss": 0.1425,
+      "loss": 0.0266,
       "step": 3996
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.258372761851749e-05,
-      "loss": 0.0814,
+      "loss": 0.062,
       "step": 3997
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.258187215882735e-05,
-      "loss": 0.3042,
+      "loss": 0.0732,
       "step": 3998
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2580016699137212e-05,
-      "loss": 0.3029,
+      "loss": 0.104,
       "step": 3999
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2578161239447074e-05,
-      "loss": 0.087,
+      "loss": 0.1321,
       "step": 4000
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2576305779756936e-05,
-      "loss": 0.0853,
+      "loss": 0.1142,
       "step": 4001
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2574450320066798e-05,
-      "loss": 0.2455,
+      "loss": 0.0712,
       "step": 4002
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.257259486037666e-05,
-      "loss": 0.1883,
+      "loss": 0.141,
       "step": 4003
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.257073940068652e-05,
-      "loss": 0.0853,
+      "loss": 0.055,
       "step": 4004
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2568883940996383e-05,
-      "loss": 0.0861,
+      "loss": 0.1616,
       "step": 4005
     },
     {
       "epoch": 1.11,
       "learning_rate": 1.2567028481306245e-05,
-      "loss": 0.1346,
+      "loss": 0.1061,
       "step": 4006
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2565173021616105e-05,
-      "loss": 0.1939,
+      "loss": 0.121,
       "step": 4007
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2563317561925967e-05,
-      "loss": 0.1393,
+      "loss": 0.1369,
       "step": 4008
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.256146210223583e-05,
-      "loss": 0.0353,
+      "loss": 0.0636,
       "step": 4009
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2559606642545693e-05,
-      "loss": 0.1431,
+      "loss": 0.0545,
       "step": 4010
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2557751182855553e-05,
-      "loss": 0.0902,
+      "loss": 0.0153,
       "step": 4011
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2555895723165415e-05,
-      "loss": 0.0887,
+      "loss": 0.0238,
       "step": 4012
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2554040263475278e-05,
-      "loss": 0.2896,
+      "loss": 0.033,
       "step": 4013
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2552184803785138e-05,
-      "loss": 0.0365,
+      "loss": 0.0568,
       "step": 4014
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2550329344095e-05,
-      "loss": 0.1396,
+      "loss": 0.0187,
       "step": 4015
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2548473884404862e-05,
-      "loss": 0.1403,
+      "loss": 0.139,
       "step": 4016
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2546618424714724e-05,
-      "loss": 0.0883,
+      "loss": 0.1013,
       "step": 4017
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2544762965024586e-05,
-      "loss": 0.1941,
+      "loss": 0.1044,
       "step": 4018
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2542907505334448e-05,
-      "loss": 0.1943,
+      "loss": 0.097,
       "step": 4019
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2541052045644308e-05,
-      "loss": 0.1401,
+      "loss": 0.0682,
       "step": 4020
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2539196585954172e-05,
-      "loss": 0.0322,
+      "loss": 0.2452,
       "step": 4021
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2537341126264033e-05,
-      "loss": 0.1897,
+      "loss": 0.0163,
       "step": 4022
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2535485666573894e-05,
-      "loss": 0.2456,
+      "loss": 0.0614,
       "step": 4023
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2533630206883755e-05,
-      "loss": 0.1397,
+      "loss": 0.1819,
       "step": 4024
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2531774747193619e-05,
-      "loss": 0.087,
+      "loss": 0.0883,
       "step": 4025
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2529919287503481e-05,
-      "loss": 0.14,
+      "loss": 0.1542,
       "step": 4026
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2528063827813341e-05,
-      "loss": 0.3006,
+      "loss": 0.0239,
       "step": 4027
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2526208368123203e-05,
-      "loss": 0.3029,
+      "loss": 0.094,
       "step": 4028
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2524352908433067e-05,
-      "loss": 0.0884,
+      "loss": 0.0526,
       "step": 4029
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2522497448742927e-05,
-      "loss": 0.0343,
+      "loss": 0.0236,
       "step": 4030
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2520641989052789e-05,
-      "loss": 0.0888,
+      "loss": 0.0184,
       "step": 4031
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.251878652936265e-05,
-      "loss": 0.1869,
+      "loss": 0.0161,
       "step": 4032
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2516931069672514e-05,
-      "loss": 0.0343,
+      "loss": 0.2993,
       "step": 4033
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2515075609982374e-05,
-      "loss": 0.1412,
+      "loss": 0.0549,
       "step": 4034
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2513220150292236e-05,
-      "loss": 0.1432,
+      "loss": 0.0565,
       "step": 4035
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2511364690602096e-05,
-      "loss": 0.1413,
+      "loss": 0.0161,
       "step": 4036
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.250950923091196e-05,
-      "loss": 0.0857,
+      "loss": 0.1148,
       "step": 4037
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2507653771221822e-05,
-      "loss": 0.033,
+      "loss": 0.1414,
       "step": 4038
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2505798311531684e-05,
-      "loss": 0.0871,
+      "loss": 0.0171,
       "step": 4039
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2503942851841544e-05,
-      "loss": 0.1392,
+      "loss": 0.1172,
       "step": 4040
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2502087392151407e-05,
-      "loss": 0.1983,
+      "loss": 0.0479,
       "step": 4041
     },
     {
       "epoch": 1.12,
       "learning_rate": 1.2500231932461269e-05,
-      "loss": 0.2452,
+      "loss": 0.1148,
       "step": 4042
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.249837647277113e-05,
-      "loss": 0.2973,
+      "loss": 0.0599,
       "step": 4043
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2496521013080991e-05,
-      "loss": 0.1351,
+      "loss": 0.1059,
       "step": 4044
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2494665553390855e-05,
-      "loss": 0.2538,
+      "loss": 0.0118,
       "step": 4045
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2492810093700715e-05,
-      "loss": 0.2487,
+      "loss": 0.0911,
       "step": 4046
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2490954634010577e-05,
-      "loss": 0.1439,
+      "loss": 0.0133,
       "step": 4047
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2489099174320439e-05,
-      "loss": 0.0866,
+      "loss": 0.1291,
       "step": 4048
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2487243714630302e-05,
-      "loss": 0.0337,
+      "loss": 0.0122,
       "step": 4049
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2485388254940162e-05,
-      "loss": 0.0846,
+      "loss": 0.0992,
       "step": 4050
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2483532795250024e-05,
-      "loss": 0.0835,
+      "loss": 0.0154,
       "step": 4051
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2481677335559884e-05,
-      "loss": 0.0328,
+      "loss": 0.0668,
       "step": 4052
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2479821875869748e-05,
-      "loss": 0.1407,
+      "loss": 0.2084,
       "step": 4053
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.247796641617961e-05,
-      "loss": 0.2486,
+      "loss": 0.024,
       "step": 4054
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2476110956489472e-05,
-      "loss": 0.1967,
+      "loss": 0.017,
       "step": 4055
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2474255496799332e-05,
-      "loss": 0.1952,
+      "loss": 0.0188,
       "step": 4056
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2472400037109196e-05,
-      "loss": 0.1884,
+      "loss": 0.1048,
       "step": 4057
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2470544577419057e-05,
-      "loss": 0.0883,
+      "loss": 0.0585,
       "step": 4058
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2468689117728918e-05,
-      "loss": 0.1375,
+      "loss": 0.0167,
       "step": 4059
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.246683365803878e-05,
-      "loss": 0.087,
+      "loss": 0.2807,
       "step": 4060
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2464978198348643e-05,
-      "loss": 0.1417,
+      "loss": 0.0367,
       "step": 4061
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2463122738658505e-05,
-      "loss": 0.0846,
+      "loss": 0.0165,
       "step": 4062
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2461267278968365e-05,
-      "loss": 0.0845,
+      "loss": 0.2478,
       "step": 4063
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2459411819278227e-05,
-      "loss": 0.0845,
+      "loss": 0.0184,
       "step": 4064
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.245755635958809e-05,
-      "loss": 0.0839,
+      "loss": 0.0742,
       "step": 4065
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.245570089989795e-05,
-      "loss": 0.1424,
+      "loss": 0.1176,
       "step": 4066
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2453845440207813e-05,
-      "loss": 0.0841,
+      "loss": 0.1826,
       "step": 4067
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2451989980517674e-05,
-      "loss": 0.0313,
+      "loss": 0.0249,
       "step": 4068
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2450134520827536e-05,
-      "loss": 0.1903,
+      "loss": 0.0482,
       "step": 4069
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2448279061137398e-05,
-      "loss": 0.1405,
+      "loss": 0.0691,
       "step": 4070
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.244642360144726e-05,
-      "loss": 0.1395,
+      "loss": 0.1187,
       "step": 4071
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.244456814175712e-05,
-      "loss": 0.0844,
+      "loss": 0.0259,
       "step": 4072
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2442712682066984e-05,
-      "loss": 0.1338,
+      "loss": 0.0683,
       "step": 4073
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2440857222376846e-05,
-      "loss": 0.0829,
+      "loss": 0.1248,
       "step": 4074
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2439001762686706e-05,
-      "loss": 0.0867,
+      "loss": 0.0235,
       "step": 4075
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2437146302996568e-05,
-      "loss": 0.0862,
+      "loss": 0.136,
       "step": 4076
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2435290843306431e-05,
-      "loss": 0.0823,
+      "loss": 0.0667,
       "step": 4077
     },
     {
       "epoch": 1.13,
       "learning_rate": 1.2433435383616293e-05,
-      "loss": 0.0821,
+      "loss": 0.0595,
       "step": 4078
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2431579923926153e-05,
-      "loss": 0.1989,
+      "loss": 0.2757,
       "step": 4079
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2429724464236015e-05,
-      "loss": 0.1989,
+      "loss": 0.022,
       "step": 4080
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2427869004545879e-05,
-      "loss": 0.0265,
+      "loss": 0.1444,
       "step": 4081
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2426013544855739e-05,
-      "loss": 0.0267,
+      "loss": 0.019,
       "step": 4082
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.24241580851656e-05,
-      "loss": 0.3136,
+      "loss": 0.1694,
       "step": 4083
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2422302625475463e-05,
-      "loss": 0.1379,
+      "loss": 0.0961,
       "step": 4084
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2420447165785326e-05,
-      "loss": 0.08,
+      "loss": 0.1071,
       "step": 4085
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2418591706095186e-05,
-      "loss": 0.1411,
+      "loss": 0.1076,
       "step": 4086
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2416736246405048e-05,
-      "loss": 0.1416,
+      "loss": 0.0863,
       "step": 4087
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2414880786714908e-05,
-      "loss": 0.1377,
+      "loss": 0.1457,
       "step": 4088
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2413025327024772e-05,
-      "loss": 0.0833,
+      "loss": 0.1406,
       "step": 4089
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2411169867334634e-05,
-      "loss": 0.201,
+      "loss": 0.031,
       "step": 4090
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2409314407644496e-05,
-      "loss": 0.1388,
+      "loss": 0.1495,
       "step": 4091
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2407458947954356e-05,
-      "loss": 0.1975,
+      "loss": 0.0633,
       "step": 4092
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.240560348826422e-05,
-      "loss": 0.0845,
+      "loss": 0.0267,
       "step": 4093
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2403748028574081e-05,
-      "loss": 0.1944,
+      "loss": 0.2095,
       "step": 4094
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2401892568883941e-05,
-      "loss": 0.0269,
+      "loss": 0.0307,
       "step": 4095
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2400037109193803e-05,
-      "loss": 0.0824,
+      "loss": 0.0688,
       "step": 4096
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2398181649503667e-05,
-      "loss": 0.0279,
+      "loss": 0.0723,
       "step": 4097
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2396326189813527e-05,
-      "loss": 0.1936,
+      "loss": 0.0276,
       "step": 4098
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2394470730123389e-05,
-      "loss": 0.2559,
+      "loss": 0.0596,
       "step": 4099
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2392615270433251e-05,
-      "loss": 0.0836,
+      "loss": 0.0981,
       "step": 4100
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2390759810743111e-05,
-      "loss": 0.2516,
+      "loss": 0.0621,
       "step": 4101
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2388904351052975e-05,
-      "loss": 0.1904,
+      "loss": 0.0694,
       "step": 4102
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2387048891362836e-05,
-      "loss": 0.0841,
+      "loss": 0.0205,
       "step": 4103
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2385193431672697e-05,
-      "loss": 0.139,
+      "loss": 0.0235,
       "step": 4104
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2383337971982559e-05,
-      "loss": 0.1399,
+      "loss": 0.153,
       "step": 4105
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2381482512292422e-05,
-      "loss": 0.1984,
+      "loss": 0.158,
       "step": 4106
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2379627052602284e-05,
-      "loss": 0.0855,
+      "loss": 0.0106,
       "step": 4107
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2377771592912144e-05,
-      "loss": 0.0857,
+      "loss": 0.0111,
       "step": 4108
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2375916133222006e-05,
-      "loss": 0.2425,
+      "loss": 0.0754,
       "step": 4109
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.237406067353187e-05,
-      "loss": 0.1917,
+      "loss": 0.0498,
       "step": 4110
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.237220521384173e-05,
-      "loss": 0.0853,
+      "loss": 0.1057,
       "step": 4111
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2370349754151592e-05,
-      "loss": 0.0297,
+      "loss": 0.0917,
       "step": 4112
     },
     {
       "epoch": 1.14,
       "learning_rate": 1.2368494294461453e-05,
-      "loss": 0.2473,
+      "loss": 0.0554,
       "step": 4113
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2366638834771317e-05,
-      "loss": 0.0859,
+      "loss": 0.111,
       "step": 4114
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2364783375081177e-05,
-      "loss": 0.1929,
+      "loss": 0.0941,
       "step": 4115
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2362927915391039e-05,
-      "loss": 0.1918,
+      "loss": 0.0519,
       "step": 4116
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.23610724557009e-05,
-      "loss": 0.1902,
+      "loss": 0.0079,
       "step": 4117
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2359216996010763e-05,
-      "loss": 0.1986,
+      "loss": 0.0987,
       "step": 4118
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2357361536320625e-05,
-      "loss": 0.1916,
+      "loss": 0.1292,
       "step": 4119
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2355506076630487e-05,
-      "loss": 0.195,
+      "loss": 0.0482,
       "step": 4120
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2353650616940347e-05,
-      "loss": 0.192,
+      "loss": 0.0169,
       "step": 4121
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.235179515725021e-05,
-      "loss": 0.0882,
+      "loss": 0.0213,
       "step": 4122
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2349939697560072e-05,
-      "loss": 0.0353,
+      "loss": 0.0892,
       "step": 4123
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2348084237869932e-05,
-      "loss": 0.0342,
+      "loss": 0.0566,
       "step": 4124
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2346228778179794e-05,
-      "loss": 0.1367,
+      "loss": 0.0167,
       "step": 4125
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2344373318489658e-05,
-      "loss": 0.0351,
+      "loss": 0.0136,
       "step": 4126
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2342517858799518e-05,
-      "loss": 0.035,
+      "loss": 0.0535,
       "step": 4127
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.234066239910938e-05,
-      "loss": 0.1391,
+      "loss": 0.0102,
       "step": 4128
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2338806939419242e-05,
-      "loss": 0.0339,
+      "loss": 0.1009,
       "step": 4129
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2336951479729105e-05,
-      "loss": 0.1426,
+      "loss": 0.0161,
       "step": 4130
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2335096020038965e-05,
-      "loss": 0.1396,
+      "loss": 0.0213,
       "step": 4131
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2333240560348827e-05,
-      "loss": 0.0842,
+      "loss": 0.1543,
       "step": 4132
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2331385100658687e-05,
-      "loss": 0.0307,
+      "loss": 0.0528,
       "step": 4133
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2329529640968551e-05,
-      "loss": 0.1972,
+      "loss": 0.3578,
       "step": 4134
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2327674181278413e-05,
-      "loss": 0.249,
+      "loss": 0.0732,
       "step": 4135
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2325818721588275e-05,
-      "loss": 0.1449,
+      "loss": 0.0844,
       "step": 4136
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2323963261898135e-05,
-      "loss": 0.0871,
+      "loss": 0.0074,
       "step": 4137
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2322107802207999e-05,
-      "loss": 0.1952,
+      "loss": 0.0105,
       "step": 4138
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.232025234251786e-05,
-      "loss": 0.084,
+      "loss": 0.0065,
       "step": 4139
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.231839688282772e-05,
-      "loss": 0.1372,
+      "loss": 0.0221,
       "step": 4140
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2316541423137582e-05,
-      "loss": 0.2547,
+      "loss": 0.1892,
       "step": 4141
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2314685963447446e-05,
-      "loss": 0.243,
+      "loss": 0.0101,
       "step": 4142
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2312830503757306e-05,
-      "loss": 0.138,
+      "loss": 0.0374,
       "step": 4143
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2310975044067168e-05,
-      "loss": 0.1422,
+      "loss": 0.1085,
       "step": 4144
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.230911958437703e-05,
-      "loss": 0.0848,
+      "loss": 0.04,
       "step": 4145
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2307264124686894e-05,
-      "loss": 0.1937,
+      "loss": 0.0569,
       "step": 4146
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2305408664996754e-05,
-      "loss": 0.1909,
+      "loss": 0.0099,
       "step": 4147
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2303553205306616e-05,
-      "loss": 0.0843,
+      "loss": 0.0193,
       "step": 4148
     },
     {
       "epoch": 1.15,
       "learning_rate": 1.2301697745616476e-05,
-      "loss": 0.0834,
+      "loss": 0.0745,
       "step": 4149
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.229984228592634e-05,
-      "loss": 0.085,
+      "loss": 0.146,
       "step": 4150
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2297986826236201e-05,
-      "loss": 0.1903,
+      "loss": 0.0299,
       "step": 4151
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2296131366546063e-05,
-      "loss": 0.0859,
+      "loss": 0.0149,
       "step": 4152
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2294275906855923e-05,
-      "loss": 0.0835,
+      "loss": 0.0521,
       "step": 4153
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2292420447165787e-05,
-      "loss": 0.0843,
+      "loss": 0.0678,
       "step": 4154
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2290564987475649e-05,
-      "loss": 0.1427,
+      "loss": 0.0097,
       "step": 4155
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2288709527785509e-05,
-      "loss": 0.1927,
+      "loss": 0.0576,
       "step": 4156
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.228685406809537e-05,
-      "loss": 0.0842,
+      "loss": 0.0269,
       "step": 4157
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2284998608405234e-05,
-      "loss": 0.1943,
+      "loss": 0.1185,
       "step": 4158
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2283143148715096e-05,
-      "loss": 0.0872,
+      "loss": 0.0089,
       "step": 4159
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2281287689024956e-05,
-      "loss": 0.2461,
+      "loss": 0.1063,
       "step": 4160
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2279432229334818e-05,
-      "loss": 0.1871,
+      "loss": 0.2197,
       "step": 4161
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2277576769644682e-05,
-      "loss": 0.0306,
+      "loss": 0.1119,
       "step": 4162
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2275721309954542e-05,
-      "loss": 0.1415,
+      "loss": 0.1628,
       "step": 4163
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2273865850264404e-05,
-      "loss": 0.2521,
+      "loss": 0.1643,
       "step": 4164
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2272010390574266e-05,
-      "loss": 0.1424,
+      "loss": 0.0687,
       "step": 4165
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2270154930884128e-05,
-      "loss": 0.031,
+      "loss": 0.0442,
       "step": 4166
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.226829947119399e-05,
-      "loss": 0.0836,
+      "loss": 0.153,
       "step": 4167
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2266444011503851e-05,
-      "loss": 0.0831,
+      "loss": 0.0278,
       "step": 4168
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2264588551813711e-05,
-      "loss": 0.0872,
+      "loss": 0.0884,
       "step": 4169
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2262733092123575e-05,
-      "loss": 0.2493,
+      "loss": 0.0786,
       "step": 4170
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2260877632433437e-05,
-      "loss": 0.1931,
+      "loss": 0.0595,
       "step": 4171
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2259022172743297e-05,
-      "loss": 0.0301,
+      "loss": 0.0309,
       "step": 4172
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2257166713053159e-05,
-      "loss": 0.1393,
+      "loss": 0.0228,
       "step": 4173
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2255311253363022e-05,
-      "loss": 0.0825,
+      "loss": 0.0556,
       "step": 4174
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2253455793672884e-05,
-      "loss": 0.1942,
+      "loss": 0.1329,
       "step": 4175
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2251600333982745e-05,
-      "loss": 0.0848,
+      "loss": 0.0867,
       "step": 4176
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2249744874292606e-05,
-      "loss": 0.3051,
+      "loss": 0.0138,
       "step": 4177
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.224788941460247e-05,
-      "loss": 0.2479,
+      "loss": 0.0143,
       "step": 4178
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.224603395491233e-05,
-      "loss": 0.1897,
+      "loss": 0.0669,
       "step": 4179
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2244178495222192e-05,
-      "loss": 0.0858,
+      "loss": 0.0496,
       "step": 4180
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2242323035532054e-05,
-      "loss": 0.1927,
+      "loss": 0.1084,
       "step": 4181
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2240467575841917e-05,
-      "loss": 0.1975,
+      "loss": 0.0544,
       "step": 4182
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.2238612116151778e-05,
-      "loss": 0.2569,
+      "loss": 0.1459,
       "step": 4183
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.223675665646164e-05,
-      "loss": 0.0854,
+      "loss": 0.0424,
       "step": 4184
     },
     {
       "epoch": 1.16,
       "learning_rate": 1.22349011967715e-05,
-      "loss": 0.0825,
+      "loss": 0.064,
       "step": 4185
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2233045737081363e-05,
-      "loss": 0.0866,
+      "loss": 0.1419,
       "step": 4186
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2231190277391225e-05,
-      "loss": 0.1934,
+      "loss": 0.1729,
       "step": 4187
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2229334817701087e-05,
-      "loss": 0.1428,
+      "loss": 0.0096,
       "step": 4188
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2227479358010947e-05,
-      "loss": 0.1932,
+      "loss": 0.015,
       "step": 4189
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.222562389832081e-05,
-      "loss": 0.1356,
+      "loss": 0.0129,
       "step": 4190
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2223768438630673e-05,
-      "loss": 0.0882,
+      "loss": 0.1015,
       "step": 4191
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2221912978940533e-05,
-      "loss": 0.1946,
+      "loss": 0.1448,
       "step": 4192
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2220057519250395e-05,
-      "loss": 0.1389,
+      "loss": 0.0671,
       "step": 4193
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2218202059560258e-05,
-      "loss": 0.2408,
+      "loss": 0.1172,
       "step": 4194
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2216346599870118e-05,
-      "loss": 0.0835,
+      "loss": 0.091,
       "step": 4195
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.221449114017998e-05,
-      "loss": 0.0852,
+      "loss": 0.0219,
       "step": 4196
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2212635680489842e-05,
-      "loss": 0.1964,
+      "loss": 0.0297,
       "step": 4197
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2210780220799706e-05,
-      "loss": 0.1898,
+      "loss": 0.0858,
       "step": 4198
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2208924761109566e-05,
-      "loss": 0.1888,
+      "loss": 0.1582,
       "step": 4199
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2207069301419428e-05,
-      "loss": 0.0355,
+      "loss": 0.0567,
       "step": 4200
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2205213841729288e-05,
-      "loss": 0.0886,
+      "loss": 0.0702,
       "step": 4201
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2203358382039151e-05,
-      "loss": 0.1888,
+      "loss": 0.0687,
       "step": 4202
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2201502922349013e-05,
-      "loss": 0.2938,
+      "loss": 0.1964,
       "step": 4203
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2199647462658875e-05,
-      "loss": 0.135,
+      "loss": 0.0295,
       "step": 4204
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2197792002968735e-05,
-      "loss": 0.0914,
+      "loss": 0.0141,
       "step": 4205
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2195936543278599e-05,
-      "loss": 0.0374,
+      "loss": 0.0212,
       "step": 4206
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.219408108358846e-05,
-      "loss": 0.2406,
+      "loss": 0.0623,
       "step": 4207
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2192225623898321e-05,
-      "loss": 0.2429,
+      "loss": 0.1577,
       "step": 4208
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2190370164208183e-05,
-      "loss": 0.1414,
+      "loss": 0.0618,
       "step": 4209
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2188514704518046e-05,
-      "loss": 0.1895,
+      "loss": 0.021,
       "step": 4210
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2186659244827908e-05,
-      "loss": 0.0878,
+      "loss": 0.1423,
       "step": 4211
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2184803785137768e-05,
-      "loss": 0.0356,
+      "loss": 0.0626,
       "step": 4212
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.218294832544763e-05,
-      "loss": 0.135,
+      "loss": 0.0583,
       "step": 4213
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2181092865757494e-05,
-      "loss": 0.24,
+      "loss": 0.0171,
       "step": 4214
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2179237406067354e-05,
-      "loss": 0.0351,
+      "loss": 0.0624,
       "step": 4215
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2177381946377216e-05,
-      "loss": 0.1428,
+      "loss": 0.1192,
       "step": 4216
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2175526486687078e-05,
-      "loss": 0.0861,
+      "loss": 0.0671,
       "step": 4217
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.217367102699694e-05,
-      "loss": 0.0343,
+      "loss": 0.1221,
       "step": 4218
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2171815567306802e-05,
-      "loss": 0.0876,
+      "loss": 0.053,
       "step": 4219
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2169960107616663e-05,
-      "loss": 0.0337,
+      "loss": 0.0594,
       "step": 4220
     },
     {
       "epoch": 1.17,
       "learning_rate": 1.2168104647926524e-05,
-      "loss": 0.0862,
+      "loss": 0.0457,
       "step": 4221
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2166249188236387e-05,
-      "loss": 0.1362,
+      "loss": 0.0517,
       "step": 4222
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2164393728546249e-05,
-      "loss": 0.1369,
+      "loss": 0.0226,
       "step": 4223
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.216253826885611e-05,
-      "loss": 0.1422,
+      "loss": 0.0604,
       "step": 4224
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2160682809165971e-05,
-      "loss": 0.1455,
+      "loss": 0.0532,
       "step": 4225
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2158827349475835e-05,
-      "loss": 0.1936,
+      "loss": 0.0119,
       "step": 4226
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2156971889785697e-05,
-      "loss": 0.0874,
+      "loss": 0.0062,
       "step": 4227
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2155116430095557e-05,
-      "loss": 0.141,
+      "loss": 0.0143,
       "step": 4228
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2153260970405419e-05,
-      "loss": 0.0829,
+      "loss": 0.0162,
       "step": 4229
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2151405510715282e-05,
-      "loss": 0.1383,
+      "loss": 0.1294,
       "step": 4230
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2149550051025142e-05,
-      "loss": 0.1392,
+      "loss": 0.0453,
       "step": 4231
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2147694591335004e-05,
-      "loss": 0.1407,
+      "loss": 0.056,
       "step": 4232
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2145839131644866e-05,
-      "loss": 0.1969,
+      "loss": 0.0575,
       "step": 4233
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.214398367195473e-05,
-      "loss": 0.1348,
+      "loss": 0.0533,
       "step": 4234
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.214212821226459e-05,
-      "loss": 0.1885,
+      "loss": 0.1601,
       "step": 4235
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2140272752574452e-05,
-      "loss": 0.0292,
+      "loss": 0.1082,
       "step": 4236
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2138417292884312e-05,
-      "loss": 0.198,
+      "loss": 0.3605,
       "step": 4237
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2136561833194175e-05,
-      "loss": 0.1948,
+      "loss": 0.0139,
       "step": 4238
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2134706373504037e-05,
-      "loss": 0.0844,
+      "loss": 0.0456,
       "step": 4239
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.21328509138139e-05,
-      "loss": 0.1902,
+      "loss": 0.0539,
       "step": 4240
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.213099545412376e-05,
-      "loss": 0.1381,
+      "loss": 0.0134,
       "step": 4241
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2129139994433623e-05,
-      "loss": 0.1401,
+      "loss": 0.0193,
       "step": 4242
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2127284534743485e-05,
-      "loss": 0.1437,
+      "loss": 0.0574,
       "step": 4243
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2125429075053345e-05,
-      "loss": 0.0849,
+      "loss": 0.0975,
       "step": 4244
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2123573615363207e-05,
-      "loss": 0.136,
+      "loss": 0.0166,
       "step": 4245
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.212171815567307e-05,
-      "loss": 0.2499,
+      "loss": 0.0124,
       "step": 4246
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.211986269598293e-05,
-      "loss": 0.1428,
+      "loss": 0.1162,
       "step": 4247
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2118007236292792e-05,
-      "loss": 0.1384,
+      "loss": 0.0487,
       "step": 4248
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2116151776602654e-05,
-      "loss": 0.0838,
+      "loss": 0.0526,
       "step": 4249
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2114296316912518e-05,
-      "loss": 0.086,
+      "loss": 0.0158,
       "step": 4250
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2112440857222378e-05,
-      "loss": 0.0842,
+      "loss": 0.0463,
       "step": 4251
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.211058539753224e-05,
-      "loss": 0.0862,
+      "loss": 0.0131,
       "step": 4252
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.21087299378421e-05,
-      "loss": 0.141,
+      "loss": 0.0546,
       "step": 4253
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2106874478151962e-05,
-      "loss": 0.0306,
+      "loss": 0.0096,
       "step": 4254
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2105019018461826e-05,
-      "loss": 0.0847,
+      "loss": 0.0461,
       "step": 4255
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2103163558771687e-05,
-      "loss": 0.2518,
+      "loss": 0.0118,
       "step": 4256
     },
     {
       "epoch": 1.18,
       "learning_rate": 1.2101308099081548e-05,
-      "loss": 0.1358,
+      "loss": 0.0371,
       "step": 4257
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.209945263939141e-05,
-      "loss": 0.0281,
+      "loss": 0.1348,
       "step": 4258
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2097597179701273e-05,
-      "loss": 0.0868,
+      "loss": 0.2103,
       "step": 4259
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2095741720011133e-05,
-      "loss": 0.1977,
+      "loss": 0.4192,
       "step": 4260
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2093886260320995e-05,
-      "loss": 0.0276,
+      "loss": 0.0148,
       "step": 4261
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2092030800630857e-05,
-      "loss": 0.2449,
+      "loss": 0.1814,
       "step": 4262
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2090175340940719e-05,
-      "loss": 0.1402,
+      "loss": 0.0541,
       "step": 4263
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.208831988125058e-05,
-      "loss": 0.1988,
+      "loss": 0.1429,
       "step": 4264
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2086464421560443e-05,
-      "loss": 0.0841,
+      "loss": 0.0677,
       "step": 4265
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2084608961870303e-05,
-      "loss": 0.1412,
+      "loss": 0.1037,
       "step": 4266
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2082753502180166e-05,
-      "loss": 0.1952,
+      "loss": 0.0567,
       "step": 4267
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2080898042490028e-05,
-      "loss": 0.0289,
+      "loss": 0.054,
       "step": 4268
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2079042582799888e-05,
-      "loss": 0.0879,
+      "loss": 0.1316,
       "step": 4269
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.207718712310975e-05,
-      "loss": 0.0286,
+      "loss": 0.0613,
       "step": 4270
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2075331663419614e-05,
-      "loss": 0.1394,
+      "loss": 0.0321,
       "step": 4271
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2073476203729476e-05,
-      "loss": 0.1412,
+      "loss": 0.0347,
       "step": 4272
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2071620744039336e-05,
-      "loss": 0.1428,
+      "loss": 0.1082,
       "step": 4273
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2069765284349198e-05,
-      "loss": 0.0813,
+      "loss": 0.0366,
       "step": 4274
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2067909824659061e-05,
-      "loss": 0.0859,
+      "loss": 0.0513,
       "step": 4275
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2066054364968921e-05,
-      "loss": 0.0842,
+      "loss": 0.0715,
       "step": 4276
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2064198905278783e-05,
-      "loss": 0.2537,
+      "loss": 0.0795,
       "step": 4277
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2062343445588645e-05,
-      "loss": 0.3062,
+      "loss": 0.0608,
       "step": 4278
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2060487985898509e-05,
-      "loss": 0.0828,
+      "loss": 0.0426,
       "step": 4279
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2058632526208369e-05,
-      "loss": 0.1382,
+      "loss": 0.1172,
       "step": 4280
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.205677706651823e-05,
-      "loss": 0.1437,
+      "loss": 0.0513,
       "step": 4281
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2054921606828091e-05,
-      "loss": 0.0838,
+      "loss": 0.0107,
       "step": 4282
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2053066147137955e-05,
-      "loss": 0.1425,
+      "loss": 0.1859,
       "step": 4283
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2051210687447816e-05,
-      "loss": 0.191,
+      "loss": 0.1453,
       "step": 4284
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2049355227757678e-05,
-      "loss": 0.3106,
+      "loss": 0.096,
       "step": 4285
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2047499768067538e-05,
-      "loss": 0.0855,
+      "loss": 0.0133,
       "step": 4286
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2045644308377402e-05,
-      "loss": 0.0844,
+      "loss": 0.0105,
       "step": 4287
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2043788848687264e-05,
-      "loss": 0.1961,
+      "loss": 0.1177,
       "step": 4288
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2041933388997124e-05,
-      "loss": 0.0293,
+      "loss": 0.0767,
       "step": 4289
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2040077929306986e-05,
-      "loss": 0.0854,
+      "loss": 0.0953,
       "step": 4290
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.203822246961685e-05,
-      "loss": 0.0816,
+      "loss": 0.0595,
       "step": 4291
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.203636700992671e-05,
-      "loss": 0.0872,
+      "loss": 0.1106,
       "step": 4292
     },
     {
       "epoch": 1.19,
       "learning_rate": 1.2034511550236572e-05,
-      "loss": 0.0292,
+      "loss": 0.233,
       "step": 4293
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2032656090546433e-05,
-      "loss": 0.1951,
+      "loss": 0.1181,
       "step": 4294
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2030800630856297e-05,
-      "loss": 0.1971,
+      "loss": 0.1486,
       "step": 4295
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2028945171166157e-05,
-      "loss": 0.1375,
+      "loss": 0.1087,
       "step": 4296
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2027089711476019e-05,
-      "loss": 0.188,
+      "loss": 0.1937,
       "step": 4297
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.202523425178588e-05,
-      "loss": 0.1427,
+      "loss": 0.2813,
       "step": 4298
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2023378792095743e-05,
-      "loss": 0.14,
+      "loss": 0.0653,
       "step": 4299
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2021523332405605e-05,
-      "loss": 0.1946,
+      "loss": 0.0216,
       "step": 4300
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2019667872715467e-05,
-      "loss": 0.193,
+      "loss": 0.1392,
       "step": 4301
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2017812413025327e-05,
-      "loss": 0.0309,
+      "loss": 0.0349,
       "step": 4302
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.201595695333519e-05,
-      "loss": 0.3601,
+      "loss": 0.067,
       "step": 4303
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2014101493645052e-05,
-      "loss": 0.0824,
+      "loss": 0.0757,
       "step": 4304
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2012246033954912e-05,
-      "loss": 0.197,
+      "loss": 0.0273,
       "step": 4305
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2010390574264774e-05,
-      "loss": 0.0856,
+      "loss": 0.0684,
       "step": 4306
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2008535114574638e-05,
-      "loss": 0.1406,
+      "loss": 0.0278,
       "step": 4307
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.20066796548845e-05,
-      "loss": 0.0329,
+      "loss": 0.16,
       "step": 4308
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.200482419519436e-05,
-      "loss": 0.1389,
+      "loss": 0.0232,
       "step": 4309
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2002968735504222e-05,
-      "loss": 0.1387,
+      "loss": 0.1415,
       "step": 4310
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.2001113275814085e-05,
-      "loss": 0.1897,
+      "loss": 0.0629,
       "step": 4311
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1999257816123945e-05,
-      "loss": 0.087,
+      "loss": 0.0572,
       "step": 4312
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1997402356433807e-05,
-      "loss": 0.1931,
+      "loss": 0.0224,
       "step": 4313
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1995546896743669e-05,
-      "loss": 0.1937,
+      "loss": 0.1585,
       "step": 4314
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1993691437053531e-05,
-      "loss": 0.1409,
+      "loss": 0.0204,
       "step": 4315
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1991835977363393e-05,
-      "loss": 0.0877,
+      "loss": 0.129,
       "step": 4316
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1989980517673255e-05,
-      "loss": 0.0852,
+      "loss": 0.0584,
       "step": 4317
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1988125057983115e-05,
-      "loss": 0.1397,
+      "loss": 0.0467,
       "step": 4318
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1986269598292978e-05,
-      "loss": 0.1368,
+      "loss": 0.0237,
       "step": 4319
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.198441413860284e-05,
-      "loss": 0.2493,
+      "loss": 0.0615,
       "step": 4320
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.19825586789127e-05,
-      "loss": 0.1441,
+      "loss": 0.057,
       "step": 4321
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1980703219222562e-05,
-      "loss": 0.0863,
+      "loss": 0.0163,
       "step": 4322
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1978847759532426e-05,
-      "loss": 0.0332,
+      "loss": 0.0841,
       "step": 4323
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1976992299842288e-05,
-      "loss": 0.1411,
+      "loss": 0.0152,
       "step": 4324
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1975136840152148e-05,
-      "loss": 0.0882,
+      "loss": 0.1546,
       "step": 4325
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.197328138046201e-05,
-      "loss": 0.0896,
+      "loss": 0.1109,
       "step": 4326
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1971425920771873e-05,
-      "loss": 0.0315,
+      "loss": 0.0653,
       "step": 4327
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1969570461081734e-05,
-      "loss": 0.2466,
+      "loss": 0.1584,
       "step": 4328
     },
     {
       "epoch": 1.2,
       "learning_rate": 1.1967715001391595e-05,
-      "loss": 0.0833,
+      "loss": 0.03,
       "step": 4329
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1965859541701457e-05,
-      "loss": 0.1929,
+      "loss": 0.0131,
       "step": 4330
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1964004082011321e-05,
-      "loss": 0.1448,
+      "loss": 0.0582,
       "step": 4331
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1962148622321181e-05,
-      "loss": 0.1442,
+      "loss": 0.0185,
       "step": 4332
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1960293162631043e-05,
-      "loss": 0.0299,
+      "loss": 0.0158,
       "step": 4333
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1958437702940903e-05,
-      "loss": 0.1971,
+      "loss": 0.0671,
       "step": 4334
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1956582243250767e-05,
-      "loss": 0.1961,
+      "loss": 0.1185,
       "step": 4335
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1954726783560629e-05,
-      "loss": 0.029,
+      "loss": 0.0093,
       "step": 4336
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.195287132387049e-05,
-      "loss": 0.1925,
+      "loss": 0.2443,
       "step": 4337
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.195101586418035e-05,
-      "loss": 0.0848,
+      "loss": 0.0112,
       "step": 4338
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1949160404490214e-05,
-      "loss": 0.0835,
+      "loss": 0.0718,
       "step": 4339
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1947304944800076e-05,
-      "loss": 0.0848,
+      "loss": 0.0604,
       "step": 4340
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1945449485109936e-05,
-      "loss": 0.0863,
+      "loss": 0.1391,
       "step": 4341
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1943594025419798e-05,
-      "loss": 0.199,
+      "loss": 0.0526,
       "step": 4342
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1941738565729662e-05,
-      "loss": 0.1965,
+      "loss": 0.0952,
       "step": 4343
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1939883106039522e-05,
-      "loss": 0.2515,
+      "loss": 0.0852,
       "step": 4344
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1938027646349384e-05,
-      "loss": 0.1427,
+      "loss": 0.0549,
       "step": 4345
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1936172186659246e-05,
-      "loss": 0.1411,
+      "loss": 0.0669,
       "step": 4346
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.193431672696911e-05,
-      "loss": 0.2533,
+      "loss": 0.0642,
       "step": 4347
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.193246126727897e-05,
-      "loss": 0.1936,
+      "loss": 0.1348,
       "step": 4348
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1930605807588831e-05,
-      "loss": 0.1368,
+      "loss": 0.1595,
       "step": 4349
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1928750347898691e-05,
-      "loss": 0.0301,
+      "loss": 0.0546,
       "step": 4350
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1926894888208555e-05,
-      "loss": 0.1378,
+      "loss": 0.1629,
       "step": 4351
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1925039428518417e-05,
-      "loss": 0.1933,
+      "loss": 0.0246,
       "step": 4352
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1923183968828279e-05,
-      "loss": 0.031,
+      "loss": 0.0307,
       "step": 4353
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1921328509138139e-05,
-      "loss": 0.1448,
+      "loss": 0.0395,
       "step": 4354
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1919473049448002e-05,
-      "loss": 0.1966,
+      "loss": 0.0198,
       "step": 4355
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1917617589757864e-05,
-      "loss": 0.1882,
+      "loss": 0.1671,
       "step": 4356
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1915762130067724e-05,
-      "loss": 0.1901,
+      "loss": 0.0663,
       "step": 4357
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1913906670377586e-05,
-      "loss": 0.0852,
+      "loss": 0.0733,
       "step": 4358
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.191205121068745e-05,
-      "loss": 0.1408,
+      "loss": 0.0206,
       "step": 4359
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1910195750997312e-05,
-      "loss": 0.0847,
+      "loss": 0.0395,
       "step": 4360
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1908340291307172e-05,
-      "loss": 0.1385,
+      "loss": 0.0601,
       "step": 4361
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1906484831617034e-05,
-      "loss": 0.0342,
+      "loss": 0.2027,
       "step": 4362
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1904629371926897e-05,
-      "loss": 0.0855,
+      "loss": 0.0606,
       "step": 4363
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.1902773912236758e-05,
-      "loss": 0.0871,
+      "loss": 0.1564,
       "step": 4364
     },
     {
       "epoch": 1.21,
       "learning_rate": 1.190091845254662e-05,
-      "loss": 0.1381,
+      "loss": 0.0292,
       "step": 4365
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1899062992856481e-05,
-      "loss": 0.0319,
+      "loss": 0.0178,
       "step": 4366
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1897207533166343e-05,
-      "loss": 0.2449,
+      "loss": 0.0178,
       "step": 4367
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1895352073476205e-05,
-      "loss": 0.1334,
+      "loss": 0.0462,
       "step": 4368
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1893496613786067e-05,
-      "loss": 0.1921,
+      "loss": 0.1457,
       "step": 4369
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1891641154095927e-05,
-      "loss": 0.0852,
+      "loss": 0.0917,
       "step": 4370
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.188978569440579e-05,
-      "loss": 0.1372,
+      "loss": 0.1203,
       "step": 4371
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1887930234715653e-05,
-      "loss": 0.1923,
+      "loss": 0.0233,
       "step": 4372
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1886074775025513e-05,
-      "loss": 0.1408,
+      "loss": 0.026,
       "step": 4373
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1884219315335375e-05,
-      "loss": 0.0852,
+      "loss": 0.1146,
       "step": 4374
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1882363855645238e-05,
-      "loss": 0.0861,
+      "loss": 0.0163,
       "step": 4375
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.18805083959551e-05,
-      "loss": 0.1391,
+      "loss": 0.1568,
       "step": 4376
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.187865293626496e-05,
-      "loss": 0.0853,
+      "loss": 0.0187,
       "step": 4377
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1876797476574822e-05,
-      "loss": 0.1375,
+      "loss": 0.1023,
       "step": 4378
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1874942016884686e-05,
-      "loss": 0.2417,
+      "loss": 0.0967,
       "step": 4379
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1873086557194546e-05,
-      "loss": 0.1961,
+      "loss": 0.0135,
       "step": 4380
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1871231097504408e-05,
-      "loss": 0.0841,
+      "loss": 0.0486,
       "step": 4381
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.186937563781427e-05,
-      "loss": 0.0844,
+      "loss": 0.1653,
       "step": 4382
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1867520178124131e-05,
-      "loss": 0.1414,
+      "loss": 0.1082,
       "step": 4383
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1865664718433993e-05,
-      "loss": 0.1366,
+      "loss": 0.0775,
       "step": 4384
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1863809258743855e-05,
-      "loss": 0.1931,
+      "loss": 0.0638,
       "step": 4385
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1861953799053715e-05,
-      "loss": 0.1417,
+      "loss": 0.0953,
       "step": 4386
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1860098339363579e-05,
-      "loss": 0.4096,
+      "loss": 0.0169,
       "step": 4387
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.185824287967344e-05,
-      "loss": 0.0864,
+      "loss": 0.0595,
       "step": 4388
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1856387419983301e-05,
-      "loss": 0.1379,
+      "loss": 0.0874,
       "step": 4389
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1854531960293163e-05,
-      "loss": 0.1906,
+      "loss": 0.0152,
       "step": 4390
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1852676500603026e-05,
-      "loss": 0.0871,
+      "loss": 0.1005,
       "step": 4391
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1850821040912888e-05,
-      "loss": 0.0891,
+      "loss": 0.012,
       "step": 4392
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1848965581222748e-05,
-      "loss": 0.1438,
+      "loss": 0.1553,
       "step": 4393
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.184711012153261e-05,
-      "loss": 0.086,
+      "loss": 0.078,
       "step": 4394
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1845254661842474e-05,
-      "loss": 0.1439,
+      "loss": 0.0642,
       "step": 4395
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1843399202152334e-05,
-      "loss": 0.0861,
+      "loss": 0.0648,
       "step": 4396
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1841543742462196e-05,
-      "loss": 0.138,
+      "loss": 0.0195,
       "step": 4397
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1839688282772058e-05,
-      "loss": 0.1949,
+      "loss": 0.1026,
       "step": 4398
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1837832823081921e-05,
-      "loss": 0.0847,
+      "loss": 0.059,
       "step": 4399
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1835977363391782e-05,
-      "loss": 0.0309,
+      "loss": 0.1293,
       "step": 4400
     },
     {
       "epoch": 1.22,
       "learning_rate": 1.1834121903701643e-05,
-      "loss": 0.1879,
+      "loss": 0.0359,
       "step": 4401
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1832266444011504e-05,
-      "loss": 0.1909,
+      "loss": 0.0593,
       "step": 4402
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1830410984321365e-05,
-      "loss": 0.1939,
+      "loss": 0.0389,
       "step": 4403
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1828555524631229e-05,
-      "loss": 0.2497,
+      "loss": 0.0876,
       "step": 4404
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1826700064941091e-05,
-      "loss": 0.1964,
+      "loss": 0.0196,
       "step": 4405
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1824844605250951e-05,
-      "loss": 0.1422,
+      "loss": 0.1049,
       "step": 4406
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1822989145560813e-05,
-      "loss": 0.302,
+      "loss": 0.0812,
       "step": 4407
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1821133685870676e-05,
-      "loss": 0.0322,
+      "loss": 0.0926,
       "step": 4408
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1819278226180537e-05,
-      "loss": 0.1373,
+      "loss": 0.0621,
       "step": 4409
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1817422766490399e-05,
-      "loss": 0.1372,
+      "loss": 0.1923,
       "step": 4410
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.181556730680026e-05,
-      "loss": 0.192,
+      "loss": 0.1184,
       "step": 4411
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1813711847110122e-05,
-      "loss": 0.243,
+      "loss": 0.1847,
       "step": 4412
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1811856387419984e-05,
-      "loss": 0.2438,
+      "loss": 0.0907,
       "step": 4413
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1810000927729846e-05,
-      "loss": 0.1377,
+      "loss": 0.2125,
       "step": 4414
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1808145468039706e-05,
-      "loss": 0.2903,
+      "loss": 0.0722,
       "step": 4415
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.180629000834957e-05,
-      "loss": 0.1958,
+      "loss": 0.1982,
       "step": 4416
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1804434548659432e-05,
-      "loss": 0.0367,
+      "loss": 0.0892,
       "step": 4417
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1802579088969292e-05,
-      "loss": 0.089,
+      "loss": 0.0546,
       "step": 4418
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1800723629279154e-05,
-      "loss": 0.0389,
+      "loss": 0.1011,
       "step": 4419
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1798868169589017e-05,
-      "loss": 0.1408,
+      "loss": 0.0458,
       "step": 4420
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1797012709898879e-05,
-      "loss": 0.0915,
+      "loss": 0.119,
       "step": 4421
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.179515725020874e-05,
-      "loss": 0.1846,
+      "loss": 0.0847,
       "step": 4422
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1793301790518601e-05,
-      "loss": 0.0865,
+      "loss": 0.0574,
       "step": 4423
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1791446330828465e-05,
-      "loss": 0.0372,
+      "loss": 0.1149,
       "step": 4424
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1789590871138325e-05,
-      "loss": 0.136,
+      "loss": 0.0617,
       "step": 4425
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1787735411448187e-05,
-      "loss": 0.1425,
+      "loss": 0.0772,
       "step": 4426
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1785879951758049e-05,
-      "loss": 0.1852,
+      "loss": 0.0883,
       "step": 4427
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1784024492067912e-05,
-      "loss": 0.2451,
+      "loss": 0.0539,
       "step": 4428
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1782169032377772e-05,
-      "loss": 0.1372,
+      "loss": 0.0625,
       "step": 4429
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1780313572687634e-05,
-      "loss": 0.0344,
+      "loss": 0.1702,
       "step": 4430
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1778458112997494e-05,
-      "loss": 0.1427,
+      "loss": 0.1112,
       "step": 4431
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1776602653307358e-05,
-      "loss": 0.2418,
+      "loss": 0.0623,
       "step": 4432
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.177474719361722e-05,
-      "loss": 0.0855,
+      "loss": 0.0671,
       "step": 4433
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1772891733927082e-05,
-      "loss": 0.1916,
+      "loss": 0.1256,
       "step": 4434
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1771036274236942e-05,
-      "loss": 0.1404,
+      "loss": 0.0832,
       "step": 4435
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1769180814546805e-05,
-      "loss": 0.1376,
+      "loss": 0.0724,
       "step": 4436
     },
     {
       "epoch": 1.23,
       "learning_rate": 1.1767325354856667e-05,
-      "loss": 0.033,
+      "loss": 0.0591,
       "step": 4437
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1765469895166528e-05,
-      "loss": 0.1402,
+      "loss": 0.1747,
       "step": 4438
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.176361443547639e-05,
-      "loss": 0.0333,
+      "loss": 0.0165,
       "step": 4439
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1761758975786253e-05,
-      "loss": 0.0335,
+      "loss": 0.1088,
       "step": 4440
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1759903516096113e-05,
-      "loss": 0.0826,
+      "loss": 0.0896,
       "step": 4441
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1758048056405975e-05,
-      "loss": 0.3081,
+      "loss": 0.0215,
       "step": 4442
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1756192596715837e-05,
-      "loss": 0.1391,
+      "loss": 0.1401,
       "step": 4443
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.17543371370257e-05,
-      "loss": 0.0312,
+      "loss": 0.113,
       "step": 4444
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.175248167733556e-05,
-      "loss": 0.3008,
+      "loss": 0.1239,
       "step": 4445
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1750626217645422e-05,
-      "loss": 0.0864,
+      "loss": 0.0181,
       "step": 4446
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1748770757955283e-05,
-      "loss": 0.2517,
+      "loss": 0.1391,
       "step": 4447
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1746915298265146e-05,
-      "loss": 0.1412,
+      "loss": 0.0879,
       "step": 4448
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1745059838575008e-05,
-      "loss": 0.084,
+      "loss": 0.0518,
       "step": 4449
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.174320437888487e-05,
-      "loss": 0.1389,
+      "loss": 0.0632,
       "step": 4450
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.174134891919473e-05,
-      "loss": 0.1403,
+      "loss": 0.1977,
       "step": 4451
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1739493459504594e-05,
-      "loss": 0.0297,
+      "loss": 0.0252,
       "step": 4452
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1737637999814456e-05,
-      "loss": 0.1404,
+      "loss": 0.1041,
       "step": 4453
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1735782540124316e-05,
-      "loss": 0.0298,
+      "loss": 0.0519,
       "step": 4454
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1733927080434178e-05,
-      "loss": 0.0838,
+      "loss": 0.0794,
       "step": 4455
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1732071620744041e-05,
-      "loss": 0.2526,
+      "loss": 0.0169,
       "step": 4456
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1730216161053903e-05,
-      "loss": 0.1934,
+      "loss": 0.0286,
       "step": 4457
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1728360701363763e-05,
-      "loss": 0.194,
+      "loss": 0.0526,
       "step": 4458
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1726505241673625e-05,
-      "loss": 0.24,
+      "loss": 0.1058,
       "step": 4459
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1724649781983489e-05,
-      "loss": 0.138,
+      "loss": 0.1403,
       "step": 4460
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1722794322293349e-05,
-      "loss": 0.1437,
+      "loss": 0.15,
       "step": 4461
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.172093886260321e-05,
-      "loss": 0.1395,
+      "loss": 0.0616,
       "step": 4462
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1719083402913073e-05,
-      "loss": 0.0844,
+      "loss": 0.109,
       "step": 4463
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1717227943222934e-05,
-      "loss": 0.1347,
+      "loss": 0.1751,
       "step": 4464
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1715372483532796e-05,
-      "loss": 0.1344,
+      "loss": 0.1139,
       "step": 4465
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1713517023842658e-05,
-      "loss": 0.2458,
+      "loss": 0.0192,
       "step": 4466
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1711661564152518e-05,
-      "loss": 0.0331,
+      "loss": 0.0289,
       "step": 4467
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1709806104462382e-05,
-      "loss": 0.2478,
+      "loss": 0.0324,
       "step": 4468
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1707950644772244e-05,
-      "loss": 0.0852,
+      "loss": 0.025,
       "step": 4469
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1706095185082104e-05,
-      "loss": 0.1378,
+      "loss": 0.0839,
       "step": 4470
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1704239725391966e-05,
-      "loss": 0.1924,
+      "loss": 0.0482,
       "step": 4471
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.170238426570183e-05,
-      "loss": 0.1407,
+      "loss": 0.1907,
       "step": 4472
     },
     {
       "epoch": 1.24,
       "learning_rate": 1.1700528806011691e-05,
-      "loss": 0.2494,
+      "loss": 0.0672,
       "step": 4473
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1698673346321551e-05,
-      "loss": 0.0826,
+      "loss": 0.0769,
       "step": 4474
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1696817886631413e-05,
-      "loss": 0.2934,
+      "loss": 0.069,
       "step": 4475
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1694962426941277e-05,
-      "loss": 0.142,
+      "loss": 0.0695,
       "step": 4476
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1693106967251137e-05,
-      "loss": 0.3011,
+      "loss": 0.0198,
       "step": 4477
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1691251507560999e-05,
-      "loss": 0.1916,
+      "loss": 0.0147,
       "step": 4478
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.168939604787086e-05,
-      "loss": 0.2443,
+      "loss": 0.0149,
       "step": 4479
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1687540588180724e-05,
-      "loss": 0.1378,
+      "loss": 0.0152,
       "step": 4480
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1685685128490585e-05,
-      "loss": 0.0864,
+      "loss": 0.1051,
       "step": 4481
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1683829668800446e-05,
-      "loss": 0.0884,
+      "loss": 0.0117,
       "step": 4482
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1681974209110307e-05,
-      "loss": 0.0389,
+      "loss": 0.0824,
       "step": 4483
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.168011874942017e-05,
-      "loss": 0.14,
+      "loss": 0.161,
       "step": 4484
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1678263289730032e-05,
-      "loss": 0.1926,
+      "loss": 0.1752,
       "step": 4485
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1676407830039894e-05,
-      "loss": 0.0854,
+      "loss": 0.1227,
       "step": 4486
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1674552370349754e-05,
-      "loss": 0.0873,
+      "loss": 0.0667,
       "step": 4487
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1672696910659618e-05,
-      "loss": 0.189,
+      "loss": 0.0926,
       "step": 4488
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.167084145096948e-05,
-      "loss": 0.0911,
+      "loss": 0.1193,
       "step": 4489
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.166898599127934e-05,
-      "loss": 0.1432,
+      "loss": 0.1643,
       "step": 4490
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1667130531589202e-05,
-      "loss": 0.1953,
+      "loss": 0.0126,
       "step": 4491
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1665275071899065e-05,
-      "loss": 0.0865,
+      "loss": 0.2928,
       "step": 4492
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1663419612208925e-05,
-      "loss": 0.0879,
+      "loss": 0.0219,
       "step": 4493
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1661564152518787e-05,
-      "loss": 0.1886,
+      "loss": 0.0882,
       "step": 4494
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1659708692828649e-05,
-      "loss": 0.2429,
+      "loss": 0.1279,
       "step": 4495
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1657853233138513e-05,
-      "loss": 0.1431,
+      "loss": 0.019,
       "step": 4496
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1655997773448373e-05,
-      "loss": 0.1419,
+      "loss": 0.0615,
       "step": 4497
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1654142313758235e-05,
-      "loss": 0.1399,
+      "loss": 0.0657,
       "step": 4498
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1652286854068095e-05,
-      "loss": 0.1359,
+      "loss": 0.0237,
       "step": 4499
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1650431394377958e-05,
-      "loss": 0.1379,
+      "loss": 0.0733,
       "step": 4500
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.164857593468782e-05,
-      "loss": 0.1407,
+      "loss": 0.075,
       "step": 4501
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1646720474997682e-05,
-      "loss": 0.187,
+      "loss": 0.0712,
       "step": 4502
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1644865015307542e-05,
-      "loss": 0.0875,
+      "loss": 0.0221,
       "step": 4503
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1643009555617406e-05,
-      "loss": 0.2394,
+      "loss": 0.0171,
       "step": 4504
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1641154095927268e-05,
-      "loss": 0.1381,
+      "loss": 0.1055,
       "step": 4505
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1639298636237128e-05,
-      "loss": 0.0365,
+      "loss": 0.1147,
       "step": 4506
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.163744317654699e-05,
-      "loss": 0.1402,
+      "loss": 0.0896,
       "step": 4507
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1635587716856853e-05,
-      "loss": 0.2947,
+      "loss": 0.0209,
       "step": 4508
     },
     {
       "epoch": 1.25,
       "learning_rate": 1.1633732257166714e-05,
-      "loss": 0.1404,
+      "loss": 0.1213,
       "step": 4509
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1631876797476575e-05,
-      "loss": 0.1421,
+      "loss": 0.0752,
       "step": 4510
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1630021337786437e-05,
-      "loss": 0.0871,
+      "loss": 0.1267,
       "step": 4511
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1628165878096301e-05,
-      "loss": 0.0344,
+      "loss": 0.1158,
       "step": 4512
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1626310418406161e-05,
-      "loss": 0.0875,
+      "loss": 0.0653,
       "step": 4513
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1624454958716023e-05,
-      "loss": 0.244,
+      "loss": 0.0394,
       "step": 4514
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1622599499025883e-05,
-      "loss": 0.0337,
+      "loss": 0.1839,
       "step": 4515
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1620744039335747e-05,
-      "loss": 0.2966,
+      "loss": 0.1648,
       "step": 4516
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1618888579645609e-05,
-      "loss": 0.0864,
+      "loss": 0.0552,
       "step": 4517
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.161703311995547e-05,
-      "loss": 0.1408,
+      "loss": 0.1277,
       "step": 4518
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.161517766026533e-05,
-      "loss": 0.1925,
+      "loss": 0.016,
       "step": 4519
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1613322200575194e-05,
-      "loss": 0.3508,
+      "loss": 0.0587,
       "step": 4520
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1611466740885056e-05,
-      "loss": 0.1409,
+      "loss": 0.0621,
       "step": 4521
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1609611281194916e-05,
-      "loss": 0.1326,
+      "loss": 0.0766,
       "step": 4522
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1607755821504778e-05,
-      "loss": 0.0331,
+      "loss": 0.0766,
       "step": 4523
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1605900361814642e-05,
-      "loss": 0.0343,
+      "loss": 0.07,
       "step": 4524
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1604044902124503e-05,
-      "loss": 0.0891,
+      "loss": 0.023,
       "step": 4525
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1602189442434364e-05,
-      "loss": 0.1384,
+      "loss": 0.1281,
       "step": 4526
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1600333982744226e-05,
-      "loss": 0.1915,
+      "loss": 0.1846,
       "step": 4527
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1598478523054089e-05,
-      "loss": 0.14,
+      "loss": 0.0104,
       "step": 4528
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.159662306336395e-05,
-      "loss": 0.1346,
+      "loss": 0.0622,
       "step": 4529
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1594767603673811e-05,
-      "loss": 0.1391,
+      "loss": 0.0542,
       "step": 4530
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1592912143983673e-05,
-      "loss": 0.1386,
+      "loss": 0.2742,
       "step": 4531
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1591056684293535e-05,
-      "loss": 0.2445,
+      "loss": 0.0887,
       "step": 4532
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1589201224603397e-05,
-      "loss": 0.0319,
+      "loss": 0.0147,
       "step": 4533
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1587345764913259e-05,
-      "loss": 0.3053,
+      "loss": 0.0225,
       "step": 4534
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1585490305223119e-05,
-      "loss": 0.2459,
+      "loss": 0.2156,
       "step": 4535
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1583634845532982e-05,
-      "loss": 0.0872,
+      "loss": 0.1261,
       "step": 4536
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1581779385842844e-05,
-      "loss": 0.0344,
+      "loss": 0.021,
       "step": 4537
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1579923926152704e-05,
-      "loss": 0.0328,
+      "loss": 0.0647,
       "step": 4538
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1578068466462566e-05,
-      "loss": 0.0849,
+      "loss": 0.0255,
       "step": 4539
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.157621300677243e-05,
-      "loss": 0.1939,
+      "loss": 0.1302,
       "step": 4540
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1574357547082292e-05,
-      "loss": 0.0873,
+      "loss": 0.0585,
       "step": 4541
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1572502087392152e-05,
-      "loss": 0.1942,
+      "loss": 0.0644,
       "step": 4542
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1570646627702014e-05,
-      "loss": 0.0861,
+      "loss": 0.0872,
       "step": 4543
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1568791168011877e-05,
-      "loss": 0.2468,
+      "loss": 0.033,
       "step": 4544
     },
     {
       "epoch": 1.26,
       "learning_rate": 1.1566935708321738e-05,
-      "loss": 0.0868,
+      "loss": 0.1017,
       "step": 4545
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.15650802486316e-05,
-      "loss": 0.087,
+      "loss": 0.0976,
       "step": 4546
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1563224788941461e-05,
-      "loss": 0.0856,
+      "loss": 0.0517,
       "step": 4547
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1561369329251325e-05,
-      "loss": 0.1362,
+      "loss": 0.0235,
       "step": 4548
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1559513869561185e-05,
-      "loss": 0.0314,
+      "loss": 0.0898,
       "step": 4549
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1557658409871047e-05,
-      "loss": 0.1368,
+      "loss": 0.316,
       "step": 4550
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1555802950180907e-05,
-      "loss": 0.1971,
+      "loss": 0.0238,
       "step": 4551
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.155394749049077e-05,
-      "loss": 0.1987,
+      "loss": 0.0771,
       "step": 4552
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1552092030800632e-05,
-      "loss": 0.2502,
+      "loss": 0.0245,
       "step": 4553
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1550236571110494e-05,
-      "loss": 0.0305,
+      "loss": 0.1708,
       "step": 4554
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1548381111420355e-05,
-      "loss": 0.0836,
+      "loss": 0.0732,
       "step": 4555
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1546525651730216e-05,
-      "loss": 0.1393,
+      "loss": 0.0549,
       "step": 4556
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.154467019204008e-05,
-      "loss": 0.2509,
+      "loss": 0.1254,
       "step": 4557
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.154281473234994e-05,
-      "loss": 0.1411,
+      "loss": 0.0224,
       "step": 4558
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1540959272659802e-05,
-      "loss": 0.0837,
+      "loss": 0.2152,
       "step": 4559
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1539103812969664e-05,
-      "loss": 0.0858,
+      "loss": 0.0226,
       "step": 4560
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1537248353279526e-05,
-      "loss": 0.1393,
+      "loss": 0.0293,
       "step": 4561
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1535392893589388e-05,
-      "loss": 0.1419,
+      "loss": 0.1049,
       "step": 4562
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.153353743389925e-05,
-      "loss": 0.1424,
+      "loss": 0.023,
       "step": 4563
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.153168197420911e-05,
-      "loss": 0.2508,
+      "loss": 0.0519,
       "step": 4564
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1529826514518973e-05,
-      "loss": 0.0295,
+      "loss": 0.0706,
       "step": 4565
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1527971054828835e-05,
-      "loss": 0.0846,
+      "loss": 0.0233,
       "step": 4566
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1526115595138695e-05,
-      "loss": 0.1397,
+      "loss": 0.0583,
       "step": 4567
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1524260135448557e-05,
-      "loss": 0.0793,
+      "loss": 0.083,
       "step": 4568
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.152240467575842e-05,
-      "loss": 0.0305,
+      "loss": 0.0944,
       "step": 4569
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1520549216068283e-05,
-      "loss": 0.1364,
+      "loss": 0.0482,
       "step": 4570
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1518693756378143e-05,
-      "loss": 0.0838,
+      "loss": 0.1584,
       "step": 4571
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1516838296688005e-05,
-      "loss": 0.1356,
+      "loss": 0.1081,
       "step": 4572
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1514982836997868e-05,
-      "loss": 0.0837,
+      "loss": 0.0189,
       "step": 4573
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1513127377307728e-05,
-      "loss": 0.2518,
+      "loss": 0.014,
       "step": 4574
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.151127191761759e-05,
-      "loss": 0.0841,
+      "loss": 0.0687,
       "step": 4575
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1509416457927452e-05,
-      "loss": 0.1399,
+      "loss": 0.1832,
       "step": 4576
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1507560998237316e-05,
-      "loss": 0.1421,
+      "loss": 0.0145,
       "step": 4577
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1505705538547176e-05,
-      "loss": 0.1365,
+      "loss": 0.0181,
       "step": 4578
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1503850078857038e-05,
-      "loss": 0.0829,
+      "loss": 0.1086,
       "step": 4579
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1501994619166898e-05,
-      "loss": 0.0825,
+      "loss": 0.0163,
       "step": 4580
     },
     {
       "epoch": 1.27,
       "learning_rate": 1.1500139159476761e-05,
-      "loss": 0.0835,
+      "loss": 0.0828,
       "step": 4581
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1498283699786623e-05,
-      "loss": 0.1974,
+      "loss": 0.1509,
       "step": 4582
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1496428240096485e-05,
-      "loss": 0.1992,
+      "loss": 0.0548,
       "step": 4583
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1494572780406345e-05,
-      "loss": 0.1338,
+      "loss": 0.0764,
       "step": 4584
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1492717320716209e-05,
-      "loss": 0.2492,
+      "loss": 0.0161,
       "step": 4585
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.149086186102607e-05,
-      "loss": 0.2512,
+      "loss": 0.0999,
       "step": 4586
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1489006401335931e-05,
-      "loss": 0.1393,
+      "loss": 0.1449,
       "step": 4587
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1487150941645793e-05,
-      "loss": 0.36,
+      "loss": 0.0181,
       "step": 4588
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1485295481955656e-05,
-      "loss": 0.1921,
+      "loss": 0.1307,
       "step": 4589
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1483440022265517e-05,
-      "loss": 0.1358,
+      "loss": 0.0713,
       "step": 4590
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1481584562575378e-05,
-      "loss": 0.1896,
+      "loss": 0.0551,
       "step": 4591
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.147972910288524e-05,
-      "loss": 0.3017,
+      "loss": 0.0306,
       "step": 4592
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1477873643195104e-05,
-      "loss": 0.1944,
+      "loss": 0.1069,
       "step": 4593
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1476018183504964e-05,
-      "loss": 0.2986,
+      "loss": 0.064,
       "step": 4594
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1474162723814826e-05,
-      "loss": 0.0893,
+      "loss": 0.056,
       "step": 4595
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1472307264124686e-05,
-      "loss": 0.189,
+      "loss": 0.173,
       "step": 4596
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.147045180443455e-05,
-      "loss": 0.1408,
+      "loss": 0.0161,
       "step": 4597
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1468596344744412e-05,
-      "loss": 0.1933,
+      "loss": 0.1316,
       "step": 4598
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1466740885054273e-05,
-      "loss": 0.1371,
+      "loss": 0.1665,
       "step": 4599
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1464885425364134e-05,
-      "loss": 0.1392,
+      "loss": 0.0208,
       "step": 4600
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1463029965673997e-05,
-      "loss": 0.1415,
+      "loss": 0.1951,
       "step": 4601
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1461174505983859e-05,
-      "loss": 0.1424,
+      "loss": 0.0726,
       "step": 4602
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.145931904629372e-05,
-      "loss": 0.0417,
+      "loss": 0.0722,
       "step": 4603
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1457463586603581e-05,
-      "loss": 0.1928,
+      "loss": 0.0681,
       "step": 4604
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1455608126913445e-05,
-      "loss": 0.2893,
+      "loss": 0.02,
       "step": 4605
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1453752667223307e-05,
-      "loss": 0.0933,
+      "loss": 0.0883,
       "step": 4606
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1451897207533167e-05,
-      "loss": 0.1377,
+      "loss": 0.1398,
       "step": 4607
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1450041747843029e-05,
-      "loss": 0.1386,
+      "loss": 0.0211,
       "step": 4608
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1448186288152892e-05,
-      "loss": 0.1381,
+      "loss": 0.0833,
       "step": 4609
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1446330828462752e-05,
-      "loss": 0.1869,
+      "loss": 0.06,
       "step": 4610
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1444475368772614e-05,
-      "loss": 0.1379,
+      "loss": 0.1823,
       "step": 4611
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1442619909082474e-05,
-      "loss": 0.1417,
+      "loss": 0.2365,
       "step": 4612
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1440764449392338e-05,
-      "loss": 0.0415,
+      "loss": 0.1909,
       "step": 4613
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.14389089897022e-05,
-      "loss": 0.0926,
+      "loss": 0.0583,
       "step": 4614
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1437053530012062e-05,
-      "loss": 0.2901,
+      "loss": 0.195,
       "step": 4615
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1435198070321922e-05,
-      "loss": 0.0389,
+      "loss": 0.0515,
       "step": 4616
     },
     {
       "epoch": 1.28,
       "learning_rate": 1.1433342610631785e-05,
-      "loss": 0.089,
+      "loss": 0.0375,
       "step": 4617
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1431487150941647e-05,
-      "loss": 0.0393,
+      "loss": 0.1736,
       "step": 4618
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1429631691251507e-05,
-      "loss": 0.1855,
+      "loss": 0.0262,
       "step": 4619
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.142777623156137e-05,
-      "loss": 0.1415,
+      "loss": 0.0932,
       "step": 4620
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1425920771871233e-05,
-      "loss": 0.2378,
+      "loss": 0.0703,
       "step": 4621
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1424065312181095e-05,
-      "loss": 0.0864,
+      "loss": 0.0912,
       "step": 4622
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1422209852490955e-05,
-      "loss": 0.1949,
+      "loss": 0.0606,
       "step": 4623
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1420354392800817e-05,
-      "loss": 0.1424,
+      "loss": 0.0588,
       "step": 4624
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.141849893311068e-05,
-      "loss": 0.0337,
+      "loss": 0.0658,
       "step": 4625
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.141664347342054e-05,
-      "loss": 0.245,
+      "loss": 0.075,
       "step": 4626
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1414788013730402e-05,
-      "loss": 0.1359,
+      "loss": 0.0524,
       "step": 4627
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1412932554040264e-05,
-      "loss": 0.1866,
+      "loss": 0.0865,
       "step": 4628
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1411077094350126e-05,
-      "loss": 0.0342,
+      "loss": 0.1721,
       "step": 4629
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1409221634659988e-05,
-      "loss": 0.191,
+      "loss": 0.0237,
       "step": 4630
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.140736617496985e-05,
-      "loss": 0.1377,
+      "loss": 0.1132,
       "step": 4631
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.140551071527971e-05,
-      "loss": 0.2951,
+      "loss": 0.1076,
       "step": 4632
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1403655255589574e-05,
-      "loss": 0.0842,
+      "loss": 0.0244,
       "step": 4633
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1401799795899436e-05,
-      "loss": 0.0328,
+      "loss": 0.1644,
       "step": 4634
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1399944336209296e-05,
-      "loss": 0.0847,
+      "loss": 0.0508,
       "step": 4635
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1398088876519158e-05,
-      "loss": 0.0881,
+      "loss": 0.0138,
       "step": 4636
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1396233416829021e-05,
-      "loss": 0.1414,
+      "loss": 0.116,
       "step": 4637
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1394377957138883e-05,
-      "loss": 0.2414,
+      "loss": 0.1698,
       "step": 4638
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1392522497448743e-05,
-      "loss": 0.2476,
+      "loss": 0.0137,
       "step": 4639
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1390667037758605e-05,
-      "loss": 0.142,
+      "loss": 0.1986,
       "step": 4640
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1388811578068469e-05,
-      "loss": 0.1424,
+      "loss": 0.1462,
       "step": 4641
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1386956118378329e-05,
-      "loss": 0.0315,
+      "loss": 0.1433,
       "step": 4642
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.138510065868819e-05,
-      "loss": 0.2495,
+      "loss": 0.0265,
       "step": 4643
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1383245198998053e-05,
-      "loss": 0.3033,
+      "loss": 0.1611,
       "step": 4644
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1381389739307916e-05,
-      "loss": 0.2492,
+      "loss": 0.1128,
       "step": 4645
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1379534279617776e-05,
-      "loss": 0.0863,
+      "loss": 0.0393,
       "step": 4646
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1377678819927638e-05,
-      "loss": 0.1888,
+      "loss": 0.0315,
       "step": 4647
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1375823360237498e-05,
-      "loss": 0.1372,
+      "loss": 0.1151,
       "step": 4648
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1373967900547362e-05,
-      "loss": 0.1397,
+      "loss": 0.1402,
       "step": 4649
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1372112440857224e-05,
-      "loss": 0.1908,
+      "loss": 0.063,
       "step": 4650
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1370256981167086e-05,
-      "loss": 0.1906,
+      "loss": 0.0641,
       "step": 4651
     },
     {
       "epoch": 1.29,
       "learning_rate": 1.1368401521476946e-05,
-      "loss": 0.1851,
+      "loss": 0.0685,
       "step": 4652
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.136654606178681e-05,
-      "loss": 0.1396,
+      "loss": 0.0592,
       "step": 4653
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1364690602096671e-05,
-      "loss": 0.2394,
+      "loss": 0.0329,
       "step": 4654
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1362835142406531e-05,
-      "loss": 0.1351,
+      "loss": 0.0637,
       "step": 4655
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1360979682716393e-05,
-      "loss": 0.1918,
+      "loss": 0.1079,
       "step": 4656
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1359124223026257e-05,
-      "loss": 0.1423,
+      "loss": 0.1089,
       "step": 4657
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1357268763336117e-05,
-      "loss": 0.2404,
+      "loss": 0.0727,
       "step": 4658
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1355413303645979e-05,
-      "loss": 0.143,
+      "loss": 0.0237,
       "step": 4659
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.135355784395584e-05,
-      "loss": 0.0398,
+      "loss": 0.1066,
       "step": 4660
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1351702384265704e-05,
-      "loss": 0.0937,
+      "loss": 0.2418,
       "step": 4661
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1349846924575565e-05,
-      "loss": 0.19,
+      "loss": 0.0458,
       "step": 4662
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1347991464885426e-05,
-      "loss": 0.0879,
+      "loss": 0.0191,
       "step": 4663
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1346136005195287e-05,
-      "loss": 0.1395,
+      "loss": 0.0144,
       "step": 4664
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.134428054550515e-05,
-      "loss": 0.0891,
+      "loss": 0.022,
       "step": 4665
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1342425085815012e-05,
-      "loss": 0.0867,
+      "loss": 0.203,
       "step": 4666
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1340569626124874e-05,
-      "loss": 0.0884,
+      "loss": 0.0056,
       "step": 4667
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1338714166434734e-05,
-      "loss": 0.0369,
+      "loss": 0.1947,
       "step": 4668
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1336858706744598e-05,
-      "loss": 0.087,
+      "loss": 0.1227,
       "step": 4669
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.133500324705446e-05,
-      "loss": 0.0866,
+      "loss": 0.0158,
       "step": 4670
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.133314778736432e-05,
-      "loss": 0.1913,
+      "loss": 0.0973,
       "step": 4671
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1331292327674182e-05,
-      "loss": 0.0866,
+      "loss": 0.0572,
       "step": 4672
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1329436867984045e-05,
-      "loss": 0.1906,
+      "loss": 0.2054,
       "step": 4673
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1327581408293907e-05,
-      "loss": 0.1937,
+      "loss": 0.1078,
       "step": 4674
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1325725948603767e-05,
-      "loss": 0.0335,
+      "loss": 0.0815,
       "step": 4675
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1323870488913629e-05,
-      "loss": 0.0862,
+      "loss": 0.0246,
       "step": 4676
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1322015029223493e-05,
-      "loss": 0.1393,
+      "loss": 0.1649,
       "step": 4677
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1320159569533353e-05,
-      "loss": 0.1384,
+      "loss": 0.03,
       "step": 4678
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1318304109843215e-05,
-      "loss": 0.0829,
+      "loss": 0.1977,
       "step": 4679
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1316448650153076e-05,
-      "loss": 0.2567,
+      "loss": 0.1446,
       "step": 4680
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1314593190462938e-05,
-      "loss": 0.1969,
+      "loss": 0.0961,
       "step": 4681
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.13127377307728e-05,
-      "loss": 0.1412,
+      "loss": 0.0264,
       "step": 4682
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1310882271082662e-05,
-      "loss": 0.1923,
+      "loss": 0.0784,
       "step": 4683
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1309026811392522e-05,
-      "loss": 0.0828,
+      "loss": 0.0624,
       "step": 4684
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1307171351702386e-05,
-      "loss": 0.0867,
+      "loss": 0.083,
       "step": 4685
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1305315892012248e-05,
-      "loss": 0.1402,
+      "loss": 0.045,
       "step": 4686
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.1303460432322108e-05,
-      "loss": 0.1391,
+      "loss": 0.1207,
       "step": 4687
     },
     {
       "epoch": 1.3,
       "learning_rate": 1.130160497263197e-05,
-      "loss": 0.144,
+      "loss": 0.0571,
       "step": 4688
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1299749512941833e-05,
-      "loss": 0.1928,
+      "loss": 0.2186,
       "step": 4689
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1297894053251695e-05,
-      "loss": 0.3004,
+      "loss": 0.0942,
       "step": 4690
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1296038593561555e-05,
-      "loss": 0.2013,
+      "loss": 0.026,
       "step": 4691
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1294183133871417e-05,
-      "loss": 0.1401,
+      "loss": 0.0589,
       "step": 4692
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.129232767418128e-05,
-      "loss": 0.085,
+      "loss": 0.0703,
       "step": 4693
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1290472214491141e-05,
-      "loss": 0.1926,
+      "loss": 0.0809,
       "step": 4694
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1288616754801003e-05,
-      "loss": 0.1924,
+      "loss": 0.0848,
       "step": 4695
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1286761295110865e-05,
-      "loss": 0.0319,
+      "loss": 0.1226,
       "step": 4696
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1284905835420728e-05,
-      "loss": 0.0857,
+      "loss": 0.0418,
       "step": 4697
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1283050375730588e-05,
-      "loss": 0.031,
+      "loss": 0.0503,
       "step": 4698
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.128119491604045e-05,
-      "loss": 0.139,
+      "loss": 0.1651,
       "step": 4699
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.127933945635031e-05,
-      "loss": 0.1386,
+      "loss": 0.0149,
       "step": 4700
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1277483996660174e-05,
-      "loss": 0.0313,
+      "loss": 0.0828,
       "step": 4701
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1275628536970036e-05,
-      "loss": 0.1922,
+      "loss": 0.136,
       "step": 4702
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1273773077279898e-05,
-      "loss": 0.141,
+      "loss": 0.0546,
       "step": 4703
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1271917617589758e-05,
-      "loss": 0.1414,
+      "loss": 0.1082,
       "step": 4704
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1270062157899622e-05,
-      "loss": 0.0309,
+      "loss": 0.0119,
       "step": 4705
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1268206698209483e-05,
-      "loss": 0.0313,
+      "loss": 0.143,
       "step": 4706
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1266351238519344e-05,
-      "loss": 0.0859,
+      "loss": 0.017,
       "step": 4707
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1264495778829205e-05,
-      "loss": 0.1391,
+      "loss": 0.1063,
       "step": 4708
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1262640319139067e-05,
-      "loss": 0.0298,
+      "loss": 0.0794,
       "step": 4709
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.126078485944893e-05,
-      "loss": 0.1398,
+      "loss": 0.106,
       "step": 4710
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1258929399758791e-05,
-      "loss": 0.3104,
+      "loss": 0.0736,
       "step": 4711
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1257073940068653e-05,
-      "loss": 0.0812,
+      "loss": 0.1649,
       "step": 4712
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1255218480378513e-05,
-      "loss": 0.1916,
+      "loss": 0.0977,
       "step": 4713
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1253363020688377e-05,
-      "loss": 0.082,
+      "loss": 0.0919,
       "step": 4714
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1251507560998239e-05,
-      "loss": 0.3071,
+      "loss": 0.0218,
       "step": 4715
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1249652101308099e-05,
-      "loss": 0.0274,
+      "loss": 0.0511,
       "step": 4716
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.124779664161796e-05,
-      "loss": 0.1387,
+      "loss": 0.0704,
       "step": 4717
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1245941181927824e-05,
-      "loss": 0.14,
+      "loss": 0.0524,
       "step": 4718
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1244085722237686e-05,
-      "loss": 0.1916,
+      "loss": 0.1145,
       "step": 4719
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1242230262547546e-05,
-      "loss": 0.0798,
+      "loss": 0.1355,
       "step": 4720
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1240374802857408e-05,
-      "loss": 0.0273,
+      "loss": 0.0408,
       "step": 4721
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1238519343167272e-05,
-      "loss": 0.2465,
+      "loss": 0.0506,
       "step": 4722
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1236663883477132e-05,
-      "loss": 0.1975,
+      "loss": 0.0275,
       "step": 4723
     },
     {
       "epoch": 1.31,
       "learning_rate": 1.1234808423786994e-05,
-      "loss": 0.1374,
+      "loss": 0.0204,
       "step": 4724
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1232952964096856e-05,
-      "loss": 0.1391,
+      "loss": 0.1141,
       "step": 4725
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1231097504406719e-05,
-      "loss": 0.0826,
+      "loss": 0.0434,
       "step": 4726
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.122924204471658e-05,
-      "loss": 0.0837,
+      "loss": 0.1413,
       "step": 4727
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1227386585026441e-05,
-      "loss": 0.0851,
+      "loss": 0.0147,
       "step": 4728
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1225531125336301e-05,
-      "loss": 0.0839,
+      "loss": 0.06,
       "step": 4729
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1223675665646165e-05,
-      "loss": 0.0831,
+      "loss": 0.0781,
       "step": 4730
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1221820205956027e-05,
-      "loss": 0.1963,
+      "loss": 0.1213,
       "step": 4731
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1219964746265887e-05,
-      "loss": 0.2537,
+      "loss": 0.1016,
       "step": 4732
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1218109286575749e-05,
-      "loss": 0.2007,
+      "loss": 0.0198,
       "step": 4733
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1216253826885612e-05,
-      "loss": 0.1427,
+      "loss": 0.042,
       "step": 4734
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1214398367195474e-05,
-      "loss": 0.0848,
+      "loss": 0.101,
       "step": 4735
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1212542907505334e-05,
-      "loss": 0.0297,
+      "loss": 0.0146,
       "step": 4736
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1210687447815196e-05,
-      "loss": 0.2494,
+      "loss": 0.0532,
       "step": 4737
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.120883198812506e-05,
-      "loss": 0.1427,
+      "loss": 0.1389,
       "step": 4738
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.120697652843492e-05,
-      "loss": 0.1389,
+      "loss": 0.046,
       "step": 4739
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1205121068744782e-05,
-      "loss": 0.0864,
+      "loss": 0.094,
       "step": 4740
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1203265609054644e-05,
-      "loss": 0.1431,
+      "loss": 0.1708,
       "step": 4741
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1201410149364507e-05,
-      "loss": 0.0843,
+      "loss": 0.0536,
       "step": 4742
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1199554689674368e-05,
-      "loss": 0.0304,
+      "loss": 0.0793,
       "step": 4743
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.119769922998423e-05,
-      "loss": 0.0849,
+      "loss": 0.0878,
       "step": 4744
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.119584377029409e-05,
-      "loss": 0.1436,
+      "loss": 0.0404,
       "step": 4745
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1193988310603953e-05,
-      "loss": 0.1387,
+      "loss": 0.061,
       "step": 4746
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1192132850913815e-05,
-      "loss": 0.1393,
+      "loss": 0.0272,
       "step": 4747
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1190277391223677e-05,
-      "loss": 0.2479,
+      "loss": 0.0966,
       "step": 4748
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1188421931533537e-05,
-      "loss": 0.0286,
+      "loss": 0.1587,
       "step": 4749
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.11865664718434e-05,
-      "loss": 0.1922,
+      "loss": 0.1043,
       "step": 4750
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1184711012153263e-05,
-      "loss": 0.2538,
+      "loss": 0.0617,
       "step": 4751
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1182855552463123e-05,
-      "loss": 0.1414,
+      "loss": 0.1811,
       "step": 4752
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1181000092772985e-05,
-      "loss": 0.0835,
+      "loss": 0.0749,
       "step": 4753
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1179144633082848e-05,
-      "loss": 0.1396,
+      "loss": 0.0897,
       "step": 4754
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1177289173392708e-05,
-      "loss": 0.1934,
+      "loss": 0.0183,
       "step": 4755
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.117543371370257e-05,
-      "loss": 0.0847,
+      "loss": 0.0591,
       "step": 4756
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1173578254012432e-05,
-      "loss": 0.1973,
+      "loss": 0.0313,
       "step": 4757
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1171722794322296e-05,
-      "loss": 0.2426,
+      "loss": 0.0713,
       "step": 4758
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1169867334632156e-05,
-      "loss": 0.029,
+      "loss": 0.105,
       "step": 4759
     },
     {
       "epoch": 1.32,
       "learning_rate": 1.1168011874942018e-05,
-      "loss": 0.192,
+      "loss": 0.0728,
       "step": 4760
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1166156415251878e-05,
-      "loss": 0.3038,
+      "loss": 0.0813,
       "step": 4761
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1164300955561741e-05,
-      "loss": 0.0869,
+      "loss": 0.0113,
       "step": 4762
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1162445495871603e-05,
-      "loss": 0.1912,
+      "loss": 0.1416,
       "step": 4763
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1160590036181465e-05,
-      "loss": 0.0863,
+      "loss": 0.0179,
       "step": 4764
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1158734576491325e-05,
-      "loss": 0.1385,
+      "loss": 0.0711,
       "step": 4765
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1156879116801189e-05,
-      "loss": 0.0872,
+      "loss": 0.0228,
       "step": 4766
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.115502365711105e-05,
-      "loss": 0.0315,
+      "loss": 0.207,
       "step": 4767
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1153168197420911e-05,
-      "loss": 0.1365,
+      "loss": 0.2517,
       "step": 4768
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1151312737730773e-05,
-      "loss": 0.1375,
+      "loss": 0.0452,
       "step": 4769
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1149457278040636e-05,
-      "loss": 0.1434,
+      "loss": 0.0559,
       "step": 4770
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1147601818350498e-05,
-      "loss": 0.3472,
+      "loss": 0.0878,
       "step": 4771
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1145746358660358e-05,
-      "loss": 0.085,
+      "loss": 0.0208,
       "step": 4772
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.114389089897022e-05,
-      "loss": 0.0851,
+      "loss": 0.0263,
       "step": 4773
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1142035439280084e-05,
-      "loss": 0.0858,
+      "loss": 0.0572,
       "step": 4774
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1140179979589944e-05,
-      "loss": 0.0838,
+      "loss": 0.0599,
       "step": 4775
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1138324519899806e-05,
-      "loss": 0.1394,
+      "loss": 0.171,
       "step": 4776
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1136469060209668e-05,
-      "loss": 0.0857,
+      "loss": 0.1173,
       "step": 4777
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.113461360051953e-05,
-      "loss": 0.1386,
+      "loss": 0.1857,
       "step": 4778
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1132758140829392e-05,
-      "loss": 0.0317,
+      "loss": 0.0532,
       "step": 4779
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1130902681139253e-05,
-      "loss": 0.2483,
+      "loss": 0.1831,
       "step": 4780
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1129047221449114e-05,
-      "loss": 0.0856,
+      "loss": 0.0583,
       "step": 4781
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1127191761758977e-05,
-      "loss": 0.0873,
+      "loss": 0.0909,
       "step": 4782
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1125336302068839e-05,
-      "loss": 0.1904,
+      "loss": 0.068,
       "step": 4783
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.11234808423787e-05,
-      "loss": 0.1383,
+      "loss": 0.0584,
       "step": 4784
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1121625382688561e-05,
-      "loss": 0.136,
+      "loss": 0.0911,
       "step": 4785
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1119769922998425e-05,
-      "loss": 0.0305,
+      "loss": 0.0959,
       "step": 4786
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1117914463308286e-05,
-      "loss": 0.087,
+      "loss": 0.1147,
       "step": 4787
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1116059003618147e-05,
-      "loss": 0.2482,
+      "loss": 0.0938,
       "step": 4788
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1114203543928009e-05,
-      "loss": 0.1368,
+      "loss": 0.086,
       "step": 4789
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1112348084237872e-05,
-      "loss": 0.1406,
+      "loss": 0.124,
       "step": 4790
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1110492624547732e-05,
-      "loss": 0.1947,
+      "loss": 0.0778,
       "step": 4791
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1108637164857594e-05,
-      "loss": 0.1941,
+      "loss": 0.0623,
       "step": 4792
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1106781705167456e-05,
-      "loss": 0.0302,
+      "loss": 0.0936,
       "step": 4793
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.110492624547732e-05,
-      "loss": 0.0852,
+      "loss": 0.0261,
       "step": 4794
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.110307078578718e-05,
-      "loss": 0.1365,
+      "loss": 0.0604,
       "step": 4795
     },
     {
       "epoch": 1.33,
       "learning_rate": 1.1101215326097042e-05,
-      "loss": 0.1946,
+      "loss": 0.0801,
       "step": 4796
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1099359866406902e-05,
-      "loss": 0.0841,
+      "loss": 0.0203,
       "step": 4797
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1097504406716765e-05,
-      "loss": 0.1454,
+      "loss": 0.0239,
       "step": 4798
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1095648947026627e-05,
-      "loss": 0.1391,
+      "loss": 0.0997,
       "step": 4799
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1093793487336489e-05,
-      "loss": 0.0289,
+      "loss": 0.0809,
       "step": 4800
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.109193802764635e-05,
-      "loss": 0.0291,
+      "loss": 0.0999,
       "step": 4801
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1090082567956213e-05,
-      "loss": 0.0856,
+      "loss": 0.0578,
       "step": 4802
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1088227108266075e-05,
-      "loss": 0.0847,
+      "loss": 0.0905,
       "step": 4803
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1086371648575935e-05,
-      "loss": 0.0826,
+      "loss": 0.0155,
       "step": 4804
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1084516188885797e-05,
-      "loss": 0.3143,
+      "loss": 0.1671,
       "step": 4805
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.108266072919566e-05,
-      "loss": 0.3067,
+      "loss": 0.1305,
       "step": 4806
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.108080526950552e-05,
-      "loss": 0.2573,
+      "loss": 0.1691,
       "step": 4807
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1078949809815382e-05,
-      "loss": 0.1395,
+      "loss": 0.0207,
       "step": 4808
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1077094350125244e-05,
-      "loss": 0.1914,
+      "loss": 0.1547,
       "step": 4809
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1075238890435108e-05,
-      "loss": 0.1939,
+      "loss": 0.0562,
       "step": 4810
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1073383430744968e-05,
-      "loss": 0.0863,
+      "loss": 0.1699,
       "step": 4811
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.107152797105483e-05,
-      "loss": 0.1931,
+      "loss": 0.0121,
       "step": 4812
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.106967251136469e-05,
-      "loss": 0.0838,
+      "loss": 0.1205,
       "step": 4813
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1067817051674554e-05,
-      "loss": 0.1374,
+      "loss": 0.0149,
       "step": 4814
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1065961591984415e-05,
-      "loss": 0.0843,
+      "loss": 0.2094,
       "step": 4815
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1064106132294277e-05,
-      "loss": 0.1939,
+      "loss": 0.0578,
       "step": 4816
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1062250672604138e-05,
-      "loss": 0.0849,
+      "loss": 0.0218,
       "step": 4817
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1060395212914001e-05,
-      "loss": 0.1376,
+      "loss": 0.023,
       "step": 4818
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1058539753223863e-05,
-      "loss": 0.0811,
+      "loss": 0.035,
       "step": 4819
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1056684293533723e-05,
-      "loss": 0.2497,
+      "loss": 0.0976,
       "step": 4820
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1054828833843585e-05,
-      "loss": 0.031,
+      "loss": 0.0962,
       "step": 4821
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1052973374153449e-05,
-      "loss": 0.031,
+      "loss": 0.0455,
       "step": 4822
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.105111791446331e-05,
-      "loss": 0.0847,
+      "loss": 0.0213,
       "step": 4823
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.104926245477317e-05,
-      "loss": 0.1935,
+      "loss": 0.0671,
       "step": 4824
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1047406995083032e-05,
-      "loss": 0.193,
+      "loss": 0.0182,
       "step": 4825
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1045551535392896e-05,
-      "loss": 0.2492,
+      "loss": 0.0257,
       "step": 4826
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1043696075702756e-05,
-      "loss": 0.1987,
+      "loss": 0.0117,
       "step": 4827
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1041840616012618e-05,
-      "loss": 0.0843,
+      "loss": 0.131,
       "step": 4828
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.103998515632248e-05,
-      "loss": 0.1381,
+      "loss": 0.0162,
       "step": 4829
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1038129696632342e-05,
-      "loss": 0.0878,
+      "loss": 0.0116,
       "step": 4830
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1036274236942204e-05,
-      "loss": 0.0862,
+      "loss": 0.018,
       "step": 4831
     },
     {
       "epoch": 1.34,
       "learning_rate": 1.1034418777252066e-05,
-      "loss": 0.2995,
+      "loss": 0.2101,
       "step": 4832
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1032563317561926e-05,
-      "loss": 0.138,
+      "loss": 0.2285,
       "step": 4833
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.103070785787179e-05,
-      "loss": 0.2999,
+      "loss": 0.1232,
       "step": 4834
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1028852398181651e-05,
-      "loss": 0.2966,
+      "loss": 0.1192,
       "step": 4835
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1026996938491511e-05,
-      "loss": 0.0334,
+      "loss": 0.1934,
       "step": 4836
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1025141478801373e-05,
-      "loss": 0.0339,
+      "loss": 0.0489,
       "step": 4837
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1023286019111237e-05,
-      "loss": 0.0337,
+      "loss": 0.0161,
       "step": 4838
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1021430559421099e-05,
-      "loss": 0.1418,
+      "loss": 0.0234,
       "step": 4839
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1019575099730959e-05,
-      "loss": 0.4036,
+      "loss": 0.1134,
       "step": 4840
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.101771964004082e-05,
-      "loss": 0.1372,
+      "loss": 0.0652,
       "step": 4841
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1015864180350684e-05,
-      "loss": 0.0855,
+      "loss": 0.0177,
       "step": 4842
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1014008720660544e-05,
-      "loss": 0.0342,
+      "loss": 0.0282,
       "step": 4843
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1012153260970406e-05,
-      "loss": 0.1384,
+      "loss": 0.0768,
       "step": 4844
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1010297801280268e-05,
-      "loss": 0.1905,
+      "loss": 0.1224,
       "step": 4845
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1008442341590132e-05,
-      "loss": 0.0344,
+      "loss": 0.0997,
       "step": 4846
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1006586881899992e-05,
-      "loss": 0.088,
+      "loss": 0.0262,
       "step": 4847
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1004731422209854e-05,
-      "loss": 0.0882,
+      "loss": 0.0161,
       "step": 4848
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1002875962519714e-05,
-      "loss": 0.2473,
+      "loss": 0.0293,
       "step": 4849
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.1001020502829578e-05,
-      "loss": 0.1419,
+      "loss": 0.2002,
       "step": 4850
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.099916504313944e-05,
-      "loss": 0.1929,
+      "loss": 0.0224,
       "step": 4851
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.09973095834493e-05,
-      "loss": 0.1411,
+      "loss": 0.0147,
       "step": 4852
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0995454123759161e-05,
-      "loss": 0.086,
+      "loss": 0.0205,
       "step": 4853
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0993598664069025e-05,
-      "loss": 0.2429,
+      "loss": 0.1961,
       "step": 4854
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0991743204378887e-05,
-      "loss": 0.2943,
+      "loss": 0.0218,
       "step": 4855
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0989887744688747e-05,
-      "loss": 0.1922,
+      "loss": 0.1319,
       "step": 4856
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0988032284998609e-05,
-      "loss": 0.1356,
+      "loss": 0.0184,
       "step": 4857
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0986176825308473e-05,
-      "loss": 0.1379,
+      "loss": 0.0612,
       "step": 4858
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0984321365618333e-05,
-      "loss": 0.1923,
+      "loss": 0.0631,
       "step": 4859
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0982465905928195e-05,
-      "loss": 0.0346,
+      "loss": 0.1403,
       "step": 4860
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0980610446238056e-05,
-      "loss": 0.1891,
+      "loss": 0.1237,
       "step": 4861
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0978754986547917e-05,
-      "loss": 0.1389,
+      "loss": 0.0136,
       "step": 4862
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.097689952685778e-05,
-      "loss": 0.1416,
+      "loss": 0.1976,
       "step": 4863
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0975044067167642e-05,
-      "loss": 0.0356,
+      "loss": 0.0828,
       "step": 4864
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0973188607477502e-05,
-      "loss": 0.1335,
+      "loss": 0.0108,
       "step": 4865
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0971333147787364e-05,
-      "loss": 0.0355,
+      "loss": 0.0623,
       "step": 4866
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.0969477688097228e-05,
-      "loss": 0.1404,
+      "loss": 0.0636,
       "step": 4867
     },
     {
       "epoch": 1.35,
       "learning_rate": 1.096762222840709e-05,
-      "loss": 0.1405,
+      "loss": 0.0983,
       "step": 4868
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.096576676871695e-05,
-      "loss": 0.087,
+      "loss": 0.0162,
       "step": 4869
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0963911309026812e-05,
-      "loss": 0.0343,
+      "loss": 0.0925,
       "step": 4870
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0962055849336675e-05,
-      "loss": 0.249,
+      "loss": 0.115,
       "step": 4871
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0960200389646535e-05,
-      "loss": 0.0821,
+      "loss": 0.1468,
       "step": 4872
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0958344929956397e-05,
-      "loss": 0.0884,
+      "loss": 0.0274,
       "step": 4873
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0956489470266259e-05,
-      "loss": 0.0311,
+      "loss": 0.0496,
       "step": 4874
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0954634010576121e-05,
-      "loss": 0.193,
+      "loss": 0.065,
       "step": 4875
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0952778550885983e-05,
-      "loss": 0.1395,
+      "loss": 0.0167,
       "step": 4876
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0950923091195845e-05,
-      "loss": 0.1395,
+      "loss": 0.0146,
       "step": 4877
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0949067631505705e-05,
-      "loss": 0.1379,
+      "loss": 0.0595,
       "step": 4878
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0947212171815568e-05,
-      "loss": 0.0852,
+      "loss": 0.153,
       "step": 4879
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.094535671212543e-05,
-      "loss": 0.0294,
+      "loss": 0.1683,
       "step": 4880
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.094350125243529e-05,
-      "loss": 0.0844,
+      "loss": 0.0226,
       "step": 4881
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0941645792745152e-05,
-      "loss": 0.2475,
+      "loss": 0.0212,
       "step": 4882
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0939790333055016e-05,
-      "loss": 0.0292,
+      "loss": 0.0747,
       "step": 4883
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0937934873364878e-05,
-      "loss": 0.1422,
+      "loss": 0.1229,
       "step": 4884
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0936079413674738e-05,
-      "loss": 0.1399,
+      "loss": 0.0776,
       "step": 4885
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.09342239539846e-05,
-      "loss": 0.0845,
+      "loss": 0.0202,
       "step": 4886
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0932368494294463e-05,
-      "loss": 0.0838,
+      "loss": 0.0212,
       "step": 4887
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0930513034604324e-05,
-      "loss": 0.0846,
+      "loss": 0.0161,
       "step": 4888
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0928657574914185e-05,
-      "loss": 0.1442,
+      "loss": 0.1907,
       "step": 4889
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0926802115224047e-05,
-      "loss": 0.2466,
+      "loss": 0.0181,
       "step": 4890
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0924946655533911e-05,
-      "loss": 0.0265,
+      "loss": 0.0814,
       "step": 4891
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0923091195843771e-05,
-      "loss": 0.1951,
+      "loss": 0.1337,
       "step": 4892
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0921235736153633e-05,
-      "loss": 0.1382,
+      "loss": 0.0585,
       "step": 4893
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0919380276463493e-05,
-      "loss": 0.2011,
+      "loss": 0.0511,
       "step": 4894
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0917524816773357e-05,
-      "loss": 0.1962,
+      "loss": 0.0204,
       "step": 4895
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0915669357083219e-05,
-      "loss": 0.2012,
+      "loss": 0.0575,
       "step": 4896
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.091381389739308e-05,
-      "loss": 0.2526,
+      "loss": 0.0216,
       "step": 4897
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.091195843770294e-05,
-      "loss": 0.1396,
+      "loss": 0.2863,
       "step": 4898
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0910102978012804e-05,
-      "loss": 0.4121,
+      "loss": 0.1177,
       "step": 4899
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0908247518322666e-05,
-      "loss": 0.138,
+      "loss": 0.1791,
       "step": 4900
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0906392058632526e-05,
-      "loss": 0.0285,
+      "loss": 0.0184,
       "step": 4901
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0904536598942388e-05,
-      "loss": 0.0838,
+      "loss": 0.1488,
       "step": 4902
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0902681139252252e-05,
-      "loss": 0.1411,
+      "loss": 0.0322,
       "step": 4903
     },
     {
       "epoch": 1.36,
       "learning_rate": 1.0900825679562112e-05,
-      "loss": 0.0828,
+      "loss": 0.1158,
       "step": 4904
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0898970219871974e-05,
-      "loss": 0.0848,
+      "loss": 0.0286,
       "step": 4905
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0897114760181836e-05,
-      "loss": 0.4066,
+      "loss": 0.0597,
       "step": 4906
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0895259300491699e-05,
-      "loss": 0.0861,
+      "loss": 0.0848,
       "step": 4907
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.089340384080156e-05,
-      "loss": 0.0832,
+      "loss": 0.0998,
       "step": 4908
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0891548381111421e-05,
-      "loss": 0.0833,
+      "loss": 0.0957,
       "step": 4909
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0889692921421281e-05,
-      "loss": 0.1398,
+      "loss": 0.0191,
       "step": 4910
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0887837461731145e-05,
-      "loss": 0.0306,
+      "loss": 0.0216,
       "step": 4911
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0885982002041007e-05,
-      "loss": 0.1379,
+      "loss": 0.097,
       "step": 4912
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0884126542350869e-05,
-      "loss": 0.0303,
+      "loss": 0.0547,
       "step": 4913
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0882271082660729e-05,
-      "loss": 0.0883,
+      "loss": 0.0651,
       "step": 4914
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0880415622970592e-05,
-      "loss": 0.1418,
+      "loss": 0.1234,
       "step": 4915
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0878560163280454e-05,
-      "loss": 0.084,
+      "loss": 0.0451,
       "step": 4916
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0876704703590314e-05,
-      "loss": 0.1983,
+      "loss": 0.1031,
       "step": 4917
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0874849243900176e-05,
-      "loss": 0.0867,
+      "loss": 0.0211,
       "step": 4918
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.087299378421004e-05,
-      "loss": 0.0289,
+      "loss": 0.0157,
       "step": 4919
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0871138324519902e-05,
-      "loss": 0.1362,
+      "loss": 0.0102,
       "step": 4920
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0869282864829762e-05,
-      "loss": 0.1896,
+      "loss": 0.0161,
       "step": 4921
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0867427405139624e-05,
-      "loss": 0.2546,
+      "loss": 0.0971,
       "step": 4922
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0865571945449487e-05,
-      "loss": 0.08,
+      "loss": 0.0138,
       "step": 4923
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0863716485759348e-05,
-      "loss": 0.1397,
+      "loss": 0.0134,
       "step": 4924
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.086186102606921e-05,
-      "loss": 0.0855,
+      "loss": 0.0554,
       "step": 4925
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0860005566379071e-05,
-      "loss": 0.195,
+      "loss": 0.1038,
       "step": 4926
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0858150106688933e-05,
-      "loss": 0.0292,
+      "loss": 0.089,
       "step": 4927
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0856294646998795e-05,
-      "loss": 0.1962,
+      "loss": 0.1561,
       "step": 4928
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0854439187308657e-05,
-      "loss": 0.0839,
+      "loss": 0.1773,
       "step": 4929
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0852583727618517e-05,
-      "loss": 0.0854,
+      "loss": 0.2002,
       "step": 4930
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.085072826792838e-05,
-      "loss": 0.0807,
+      "loss": 0.0885,
       "step": 4931
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0848872808238242e-05,
-      "loss": 0.1377,
+      "loss": 0.1021,
       "step": 4932
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0847017348548103e-05,
-      "loss": 0.0277,
+      "loss": 0.0144,
       "step": 4933
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0845161888857965e-05,
-      "loss": 0.2542,
+      "loss": 0.1855,
       "step": 4934
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0843306429167828e-05,
-      "loss": 0.1381,
+      "loss": 0.0549,
       "step": 4935
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.084145096947769e-05,
-      "loss": 0.3152,
+      "loss": 0.0649,
       "step": 4936
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.083959550978755e-05,
-      "loss": 0.3053,
+      "loss": 0.0167,
       "step": 4937
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0837740050097412e-05,
-      "loss": 0.0273,
+      "loss": 0.034,
       "step": 4938
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0835884590407276e-05,
-      "loss": 0.0851,
+      "loss": 0.0904,
       "step": 4939
     },
     {
       "epoch": 1.37,
       "learning_rate": 1.0834029130717136e-05,
-      "loss": 0.0292,
+      "loss": 0.0903,
       "step": 4940
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0832173671026998e-05,
-      "loss": 0.3061,
+      "loss": 0.0201,
       "step": 4941
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.083031821133686e-05,
-      "loss": 0.0829,
+      "loss": 0.0238,
       "step": 4942
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0828462751646723e-05,
-      "loss": 0.0822,
+      "loss": 0.0143,
       "step": 4943
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0826607291956583e-05,
-      "loss": 0.1342,
+      "loss": 0.0518,
       "step": 4944
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0824751832266445e-05,
-      "loss": 0.1959,
+      "loss": 0.0647,
       "step": 4945
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0822896372576305e-05,
-      "loss": 0.0307,
+      "loss": 0.0564,
       "step": 4946
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0821040912886169e-05,
-      "loss": 0.0838,
+      "loss": 0.012,
       "step": 4947
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.081918545319603e-05,
-      "loss": 0.1362,
+      "loss": 0.2018,
       "step": 4948
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0817329993505893e-05,
-      "loss": 0.0293,
+      "loss": 0.0258,
       "step": 4949
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0815474533815753e-05,
-      "loss": 0.1964,
+      "loss": 0.1049,
       "step": 4950
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0813619074125616e-05,
-      "loss": 0.1919,
+      "loss": 0.0154,
       "step": 4951
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0811763614435478e-05,
-      "loss": 0.2508,
+      "loss": 0.0526,
       "step": 4952
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0809908154745338e-05,
-      "loss": 0.141,
+      "loss": 0.1941,
       "step": 4953
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.08080526950552e-05,
-      "loss": 0.0842,
+      "loss": 0.0457,
       "step": 4954
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0806197235365064e-05,
-      "loss": 0.0848,
+      "loss": 0.0212,
       "step": 4955
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0804341775674924e-05,
-      "loss": 0.0828,
+      "loss": 0.0952,
       "step": 4956
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0802486315984786e-05,
-      "loss": 0.0298,
+      "loss": 0.01,
       "step": 4957
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0800630856294648e-05,
-      "loss": 0.1931,
+      "loss": 0.0182,
       "step": 4958
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0798775396604511e-05,
-      "loss": 0.1401,
+      "loss": 0.1983,
       "step": 4959
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0796919936914371e-05,
-      "loss": 0.1969,
+      "loss": 0.0608,
       "step": 4960
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0795064477224233e-05,
-      "loss": 0.0853,
+      "loss": 0.0175,
       "step": 4961
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0793209017534094e-05,
-      "loss": 0.0862,
+      "loss": 0.0192,
       "step": 4962
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0791353557843957e-05,
-      "loss": 0.1421,
+      "loss": 0.0493,
       "step": 4963
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0789498098153819e-05,
-      "loss": 0.1384,
+      "loss": 0.0168,
       "step": 4964
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.078764263846368e-05,
-      "loss": 0.14,
+      "loss": 0.0824,
       "step": 4965
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0785787178773541e-05,
-      "loss": 0.0881,
+      "loss": 0.0598,
       "step": 4966
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0783931719083405e-05,
-      "loss": 0.0853,
+      "loss": 0.0521,
       "step": 4967
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0782076259393266e-05,
-      "loss": 0.1912,
+      "loss": 0.151,
       "step": 4968
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0780220799703127e-05,
-      "loss": 0.1331,
+      "loss": 0.0622,
       "step": 4969
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0778365340012988e-05,
-      "loss": 0.3014,
+      "loss": 0.0557,
       "step": 4970
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0776509880322852e-05,
-      "loss": 0.1388,
+      "loss": 0.1069,
       "step": 4971
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0774654420632712e-05,
-      "loss": 0.0294,
+      "loss": 0.1071,
       "step": 4972
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0772798960942574e-05,
-      "loss": 0.14,
+      "loss": 0.0848,
       "step": 4973
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.0770943501252436e-05,
-      "loss": 0.0807,
+      "loss": 0.0221,
       "step": 4974
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.07690880415623e-05,
-      "loss": 0.0833,
+      "loss": 0.0934,
       "step": 4975
     },
     {
       "epoch": 1.38,
       "learning_rate": 1.076723258187216e-05,
-      "loss": 0.2471,
+      "loss": 0.1805,
       "step": 4976
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0765377122182022e-05,
-      "loss": 0.0826,
+      "loss": 0.0108,
       "step": 4977
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0763521662491882e-05,
-      "loss": 0.0847,
+      "loss": 0.0704,
       "step": 4978
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0761666202801745e-05,
-      "loss": 0.1376,
+      "loss": 0.028,
       "step": 4979
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0759810743111607e-05,
-      "loss": 0.2489,
+      "loss": 0.0542,
       "step": 4980
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0757955283421469e-05,
-      "loss": 0.0885,
+      "loss": 0.0616,
       "step": 4981
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.075609982373133e-05,
-      "loss": 0.1932,
+      "loss": 0.0492,
       "step": 4982
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0754244364041193e-05,
-      "loss": 0.1382,
+      "loss": 0.0418,
       "step": 4983
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0752388904351055e-05,
-      "loss": 0.1874,
+      "loss": 0.0213,
       "step": 4984
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0750533444660915e-05,
-      "loss": 0.1961,
+      "loss": 0.1262,
       "step": 4985
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0748677984970777e-05,
-      "loss": 0.3081,
+      "loss": 0.1477,
       "step": 4986
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.074682252528064e-05,
-      "loss": 0.0851,
+      "loss": 0.034,
       "step": 4987
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0744967065590502e-05,
-      "loss": 0.1387,
+      "loss": 0.0256,
       "step": 4988
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0743111605900362e-05,
-      "loss": 0.1409,
+      "loss": 0.1171,
       "step": 4989
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0741256146210224e-05,
-      "loss": 0.1408,
+      "loss": 0.0256,
       "step": 4990
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0739400686520088e-05,
-      "loss": 0.0321,
+      "loss": 0.0756,
       "step": 4991
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0737545226829948e-05,
-      "loss": 0.2513,
+      "loss": 0.0552,
       "step": 4992
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.073568976713981e-05,
-      "loss": 0.1921,
+      "loss": 0.0754,
       "step": 4993
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0733834307449672e-05,
-      "loss": 0.0875,
+      "loss": 0.1637,
       "step": 4994
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0731978847759534e-05,
-      "loss": 0.2422,
+      "loss": 0.0139,
       "step": 4995
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0730123388069395e-05,
-      "loss": 0.189,
+      "loss": 0.0689,
       "step": 4996
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0728267928379257e-05,
-      "loss": 0.1374,
+      "loss": 0.0528,
       "step": 4997
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0726412468689117e-05,
-      "loss": 0.0343,
+      "loss": 0.0171,
       "step": 4998
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0724557008998981e-05,
-      "loss": 0.0358,
+      "loss": 0.0769,
       "step": 4999
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0722701549308843e-05,
-      "loss": 0.3455,
+      "loss": 0.0203,
       "step": 5000
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0720846089618703e-05,
-      "loss": 0.0844,
+      "loss": 0.0718,
       "step": 5001
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0718990629928565e-05,
-      "loss": 0.1423,
+      "loss": 0.0849,
       "step": 5002
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0717135170238428e-05,
-      "loss": 0.0899,
+      "loss": 0.05,
       "step": 5003
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.071527971054829e-05,
-      "loss": 0.0907,
+      "loss": 0.1146,
       "step": 5004
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.071342425085815e-05,
-      "loss": 0.14,
+      "loss": 0.0801,
       "step": 5005
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0711568791168012e-05,
-      "loss": 0.1397,
+      "loss": 0.1599,
       "step": 5006
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0709713331477876e-05,
-      "loss": 0.1388,
+      "loss": 0.1072,
       "step": 5007
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0707857871787736e-05,
-      "loss": 0.0341,
+      "loss": 0.1797,
       "step": 5008
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0706002412097598e-05,
-      "loss": 0.1363,
+      "loss": 0.0777,
       "step": 5009
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.070414695240746e-05,
-      "loss": 0.1939,
+      "loss": 0.0579,
       "step": 5010
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.070229149271732e-05,
-      "loss": 0.0322,
+      "loss": 0.1137,
       "step": 5011
     },
     {
       "epoch": 1.39,
       "learning_rate": 1.0700436033027184e-05,
-      "loss": 0.1873,
+      "loss": 0.1153,
       "step": 5012
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0698580573337046e-05,
-      "loss": 0.034,
+      "loss": 0.0613,
       "step": 5013
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0696725113646906e-05,
-      "loss": 0.0322,
+      "loss": 0.0683,
       "step": 5014
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0694869653956768e-05,
-      "loss": 0.302,
+      "loss": 0.0567,
       "step": 5015
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0693014194266631e-05,
-      "loss": 0.2391,
+      "loss": 0.1227,
       "step": 5016
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0691158734576493e-05,
-      "loss": 0.132,
+      "loss": 0.1107,
       "step": 5017
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0689303274886353e-05,
-      "loss": 0.0868,
+      "loss": 0.0664,
       "step": 5018
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0687447815196215e-05,
-      "loss": 0.1942,
+      "loss": 0.0664,
       "step": 5019
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0685592355506079e-05,
-      "loss": 0.1944,
+      "loss": 0.0186,
       "step": 5020
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0683736895815939e-05,
-      "loss": 0.0877,
+      "loss": 0.0667,
       "step": 5021
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.06818814361258e-05,
-      "loss": 0.1347,
+      "loss": 0.0545,
       "step": 5022
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0680025976435663e-05,
-      "loss": 0.2494,
+      "loss": 0.0267,
       "step": 5023
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0678170516745524e-05,
-      "loss": 0.2461,
+      "loss": 0.0378,
       "step": 5024
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0676315057055386e-05,
-      "loss": 0.1417,
+      "loss": 0.0864,
       "step": 5025
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0674459597365248e-05,
-      "loss": 0.1389,
+      "loss": 0.0161,
       "step": 5026
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0672604137675108e-05,
-      "loss": 0.1389,
+      "loss": 0.0591,
       "step": 5027
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0670748677984972e-05,
-      "loss": 0.0873,
+      "loss": 0.1434,
       "step": 5028
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0668893218294834e-05,
-      "loss": 0.1914,
+      "loss": 0.0113,
       "step": 5029
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0667037758604694e-05,
-      "loss": 0.0876,
+      "loss": 0.0148,
       "step": 5030
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0665182298914556e-05,
-      "loss": 0.1926,
+      "loss": 0.0202,
       "step": 5031
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.066332683922442e-05,
-      "loss": 0.034,
+      "loss": 0.007,
       "step": 5032
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0661471379534281e-05,
-      "loss": 0.0347,
+      "loss": 0.0187,
       "step": 5033
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0659615919844141e-05,
-      "loss": 0.2449,
+      "loss": 0.0378,
       "step": 5034
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0657760460154003e-05,
-      "loss": 0.3474,
+      "loss": 0.15,
       "step": 5035
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0655905000463867e-05,
-      "loss": 0.1408,
+      "loss": 0.1468,
       "step": 5036
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0654049540773727e-05,
-      "loss": 0.1396,
+      "loss": 0.1239,
       "step": 5037
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0652194081083589e-05,
-      "loss": 0.0848,
+      "loss": 0.0213,
       "step": 5038
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.065033862139345e-05,
-      "loss": 0.1416,
+      "loss": 0.0697,
       "step": 5039
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0648483161703314e-05,
-      "loss": 0.0856,
+      "loss": 0.0142,
       "step": 5040
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0646627702013174e-05,
-      "loss": 0.086,
+      "loss": 0.1177,
       "step": 5041
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0644772242323036e-05,
-      "loss": 0.1357,
+      "loss": 0.0135,
       "step": 5042
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0642916782632897e-05,
-      "loss": 0.1367,
+      "loss": 0.0122,
       "step": 5043
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.064106132294276e-05,
-      "loss": 0.1364,
+      "loss": 0.1111,
       "step": 5044
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0639205863252622e-05,
-      "loss": 0.1438,
+      "loss": 0.0724,
       "step": 5045
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0637350403562484e-05,
-      "loss": 0.1904,
+      "loss": 0.1002,
       "step": 5046
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0635494943872344e-05,
-      "loss": 0.1386,
+      "loss": 0.1263,
       "step": 5047
     },
     {
       "epoch": 1.4,
       "learning_rate": 1.0633639484182208e-05,
-      "loss": 0.1951,
+      "loss": 0.0207,
       "step": 5048
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.063178402449207e-05,
-      "loss": 0.2436,
+      "loss": 0.0888,
       "step": 5049
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.062992856480193e-05,
-      "loss": 0.3012,
+      "loss": 0.0731,
       "step": 5050
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0628073105111792e-05,
-      "loss": 0.1907,
+      "loss": 0.0584,
       "step": 5051
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0626217645421655e-05,
-      "loss": 0.1908,
+      "loss": 0.1578,
       "step": 5052
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0624362185731515e-05,
-      "loss": 0.0361,
+      "loss": 0.1076,
       "step": 5053
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0622506726041377e-05,
-      "loss": 0.0875,
+      "loss": 0.0255,
       "step": 5054
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0620651266351239e-05,
-      "loss": 0.0901,
+      "loss": 0.0998,
       "step": 5055
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0618795806661103e-05,
-      "loss": 0.1403,
+      "loss": 0.0922,
       "step": 5056
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0616940346970963e-05,
-      "loss": 0.0364,
+      "loss": 0.1171,
       "step": 5057
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0615084887280825e-05,
-      "loss": 0.0363,
+      "loss": 0.0683,
       "step": 5058
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0613229427590685e-05,
-      "loss": 0.0343,
+      "loss": 0.0721,
       "step": 5059
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0611373967900548e-05,
-      "loss": 0.1383,
+      "loss": 0.1216,
       "step": 5060
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.060951850821041e-05,
-      "loss": 0.1372,
+      "loss": 0.0278,
       "step": 5061
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0607663048520272e-05,
-      "loss": 0.0875,
+      "loss": 0.0274,
       "step": 5062
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0605807588830132e-05,
-      "loss": 0.0834,
+      "loss": 0.095,
       "step": 5063
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0603952129139996e-05,
-      "loss": 0.1968,
+      "loss": 0.0835,
       "step": 5064
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0602096669449858e-05,
-      "loss": 0.2456,
+      "loss": 0.1529,
       "step": 5065
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0600241209759718e-05,
-      "loss": 0.0859,
+      "loss": 0.1772,
       "step": 5066
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.059838575006958e-05,
-      "loss": 0.0884,
+      "loss": 0.3128,
       "step": 5067
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0596530290379443e-05,
-      "loss": 0.0304,
+      "loss": 0.1471,
       "step": 5068
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0594674830689305e-05,
-      "loss": 0.1408,
+      "loss": 0.1388,
       "step": 5069
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0592819370999165e-05,
-      "loss": 0.251,
+      "loss": 0.027,
       "step": 5070
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0590963911309027e-05,
-      "loss": 0.1893,
+      "loss": 0.1202,
       "step": 5071
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.058910845161889e-05,
-      "loss": 0.0815,
+      "loss": 0.1147,
       "step": 5072
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0587252991928751e-05,
-      "loss": 0.1431,
+      "loss": 0.0258,
       "step": 5073
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0585397532238613e-05,
-      "loss": 0.0806,
+      "loss": 0.0686,
       "step": 5074
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0583542072548475e-05,
-      "loss": 0.2444,
+      "loss": 0.0283,
       "step": 5075
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0581686612858337e-05,
-      "loss": 0.1931,
+      "loss": 0.1446,
       "step": 5076
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0579831153168198e-05,
-      "loss": 0.0286,
+      "loss": 0.0806,
       "step": 5077
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.057797569347806e-05,
-      "loss": 0.1365,
+      "loss": 0.0641,
       "step": 5078
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.057612023378792e-05,
-      "loss": 0.1389,
+      "loss": 0.126,
       "step": 5079
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0574264774097784e-05,
-      "loss": 0.0822,
+      "loss": 0.1319,
       "step": 5080
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0572409314407646e-05,
-      "loss": 0.1396,
+      "loss": 0.0541,
       "step": 5081
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0570553854717506e-05,
-      "loss": 0.0843,
+      "loss": 0.0592,
       "step": 5082
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0568698395027368e-05,
-      "loss": 0.1408,
+      "loss": 0.1075,
       "step": 5083
     },
     {
       "epoch": 1.41,
       "learning_rate": 1.0566842935337232e-05,
-      "loss": 0.2517,
+      "loss": 0.0589,
       "step": 5084
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0564987475647093e-05,
-      "loss": 0.1848,
+      "loss": 0.0654,
       "step": 5085
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0563132015956954e-05,
-      "loss": 0.1355,
+      "loss": 0.1029,
       "step": 5086
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0561276556266815e-05,
-      "loss": 0.0299,
+      "loss": 0.1149,
       "step": 5087
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0559421096576679e-05,
-      "loss": 0.1375,
+      "loss": 0.1078,
       "step": 5088
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.055756563688654e-05,
-      "loss": 0.1382,
+      "loss": 0.0585,
       "step": 5089
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0555710177196401e-05,
-      "loss": 0.1407,
+      "loss": 0.0182,
       "step": 5090
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0553854717506263e-05,
-      "loss": 0.081,
+      "loss": 0.0543,
       "step": 5091
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0551999257816125e-05,
-      "loss": 0.0879,
+      "loss": 0.0891,
       "step": 5092
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0550143798125987e-05,
-      "loss": 0.2468,
+      "loss": 0.2187,
       "step": 5093
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0548288338435849e-05,
-      "loss": 0.0859,
+      "loss": 0.0176,
       "step": 5094
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0546432878745709e-05,
-      "loss": 0.133,
+      "loss": 0.0164,
       "step": 5095
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0544577419055572e-05,
-      "loss": 0.2474,
+      "loss": 0.2568,
       "step": 5096
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0542721959365434e-05,
-      "loss": 0.0301,
+      "loss": 0.0592,
       "step": 5097
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0540866499675294e-05,
-      "loss": 0.4111,
+      "loss": 0.1478,
       "step": 5098
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0539011039985156e-05,
-      "loss": 0.138,
+      "loss": 0.0595,
       "step": 5099
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.053715558029502e-05,
-      "loss": 0.1979,
+      "loss": 0.041,
       "step": 5100
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0535300120604882e-05,
-      "loss": 0.139,
+      "loss": 0.0799,
       "step": 5101
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0533444660914742e-05,
-      "loss": 0.1384,
+      "loss": 0.0753,
       "step": 5102
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0531589201224604e-05,
-      "loss": 0.191,
+      "loss": 0.0545,
       "step": 5103
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0529733741534467e-05,
-      "loss": 0.0838,
+      "loss": 0.0791,
       "step": 5104
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0527878281844327e-05,
-      "loss": 0.0825,
+      "loss": 0.0708,
       "step": 5105
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.052602282215419e-05,
-      "loss": 0.1943,
+      "loss": 0.0161,
       "step": 5106
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0524167362464051e-05,
-      "loss": 0.1368,
+      "loss": 0.0973,
       "step": 5107
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0522311902773915e-05,
-      "loss": 0.089,
+      "loss": 0.1053,
       "step": 5108
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0520456443083775e-05,
-      "loss": 0.1397,
+      "loss": 0.0502,
       "step": 5109
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0518600983393637e-05,
-      "loss": 0.0862,
+      "loss": 0.0328,
       "step": 5110
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0516745523703497e-05,
-      "loss": 0.0327,
+      "loss": 0.0672,
       "step": 5111
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.051489006401336e-05,
-      "loss": 0.1939,
+      "loss": 0.1255,
       "step": 5112
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0513034604323222e-05,
-      "loss": 0.1422,
+      "loss": 0.0492,
       "step": 5113
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0511179144633084e-05,
-      "loss": 0.0878,
+      "loss": 0.0111,
       "step": 5114
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0509323684942944e-05,
-      "loss": 0.1391,
+      "loss": 0.0555,
       "step": 5115
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0507468225252808e-05,
-      "loss": 0.0868,
+      "loss": 0.0116,
       "step": 5116
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.050561276556267e-05,
-      "loss": 0.3503,
+      "loss": 0.1205,
       "step": 5117
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.050375730587253e-05,
-      "loss": 0.0324,
+      "loss": 0.0564,
       "step": 5118
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0501901846182392e-05,
-      "loss": 0.0331,
+      "loss": 0.0132,
       "step": 5119
     },
     {
       "epoch": 1.42,
       "learning_rate": 1.0500046386492255e-05,
-      "loss": 0.2516,
+      "loss": 0.2016,
       "step": 5120
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0498190926802116e-05,
-      "loss": 0.0847,
+      "loss": 0.1401,
       "step": 5121
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0496335467111978e-05,
-      "loss": 0.1417,
+      "loss": 0.0733,
       "step": 5122
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.049448000742184e-05,
-      "loss": 0.0855,
+      "loss": 0.1306,
       "step": 5123
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0492624547731703e-05,
-      "loss": 0.0844,
+      "loss": 0.0541,
       "step": 5124
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0490769088041563e-05,
-      "loss": 0.1379,
+      "loss": 0.1154,
       "step": 5125
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0488913628351425e-05,
-      "loss": 0.1388,
+      "loss": 0.218,
       "step": 5126
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0487058168661285e-05,
-      "loss": 0.0855,
+      "loss": 0.0654,
       "step": 5127
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0485202708971149e-05,
-      "loss": 0.1912,
+      "loss": 0.1039,
       "step": 5128
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.048334724928101e-05,
-      "loss": 0.1918,
+      "loss": 0.0215,
       "step": 5129
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0481491789590873e-05,
-      "loss": 0.1402,
+      "loss": 0.0977,
       "step": 5130
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0479636329900733e-05,
-      "loss": 0.084,
+      "loss": 0.0927,
       "step": 5131
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0477780870210596e-05,
-      "loss": 0.03,
+      "loss": 0.0229,
       "step": 5132
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0475925410520458e-05,
-      "loss": 0.1372,
+      "loss": 0.0314,
       "step": 5133
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0474069950830318e-05,
-      "loss": 0.1365,
+      "loss": 0.0499,
       "step": 5134
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.047221449114018e-05,
-      "loss": 0.1971,
+      "loss": 0.0175,
       "step": 5135
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0470359031450044e-05,
-      "loss": 0.028,
+      "loss": 0.1299,
       "step": 5136
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0468503571759906e-05,
-      "loss": 0.1329,
+      "loss": 0.0458,
       "step": 5137
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0466648112069766e-05,
-      "loss": 0.1929,
+      "loss": 0.023,
       "step": 5138
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0464792652379628e-05,
-      "loss": 0.0837,
+      "loss": 0.0573,
       "step": 5139
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0462937192689491e-05,
-      "loss": 0.0828,
+      "loss": 0.0187,
       "step": 5140
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0461081732999351e-05,
-      "loss": 0.1363,
+      "loss": 0.0193,
       "step": 5141
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0459226273309213e-05,
-      "loss": 0.1374,
+      "loss": 0.054,
       "step": 5142
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0457370813619075e-05,
-      "loss": 0.0294,
+      "loss": 0.08,
       "step": 5143
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0455515353928937e-05,
-      "loss": 0.3091,
+      "loss": 0.0554,
       "step": 5144
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0453659894238799e-05,
-      "loss": 0.1406,
+      "loss": 0.1117,
       "step": 5145
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.045180443454866e-05,
-      "loss": 0.1406,
+      "loss": 0.0623,
       "step": 5146
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0449948974858521e-05,
-      "loss": 0.083,
+      "loss": 0.0488,
       "step": 5147
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0448093515168384e-05,
-      "loss": 0.1884,
+      "loss": 0.0116,
       "step": 5148
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0446238055478246e-05,
-      "loss": 0.0865,
+      "loss": 0.107,
       "step": 5149
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0444382595788107e-05,
-      "loss": 0.0291,
+      "loss": 0.1491,
       "step": 5150
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0442527136097968e-05,
-      "loss": 0.2023,
+      "loss": 0.2144,
       "step": 5151
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0440671676407832e-05,
-      "loss": 0.3039,
+      "loss": 0.0639,
       "step": 5152
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0438816216717694e-05,
-      "loss": 0.0824,
+      "loss": 0.1973,
       "step": 5153
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0436960757027554e-05,
-      "loss": 0.138,
+      "loss": 0.0125,
       "step": 5154
     },
     {
       "epoch": 1.43,
       "learning_rate": 1.0435105297337416e-05,
-      "loss": 0.2,
+      "loss": 0.1607,
       "step": 5155
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.043324983764728e-05,
-      "loss": 0.0286,
+      "loss": 0.0881,
       "step": 5156
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.043139437795714e-05,
-      "loss": 0.1405,
+      "loss": 0.1525,
       "step": 5157
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0429538918267001e-05,
-      "loss": 0.1963,
+      "loss": 0.0566,
       "step": 5158
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0427683458576863e-05,
-      "loss": 0.1936,
+      "loss": 0.0582,
       "step": 5159
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0425827998886727e-05,
-      "loss": 0.0831,
+      "loss": 0.0237,
       "step": 5160
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0423972539196587e-05,
-      "loss": 0.0863,
+      "loss": 0.0157,
       "step": 5161
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0422117079506449e-05,
-      "loss": 0.1929,
+      "loss": 0.0981,
       "step": 5162
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.042026161981631e-05,
-      "loss": 0.0296,
+      "loss": 0.0247,
       "step": 5163
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0418406160126171e-05,
-      "loss": 0.2456,
+      "loss": 0.0313,
       "step": 5164
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0416550700436035e-05,
-      "loss": 0.1385,
+      "loss": 0.024,
       "step": 5165
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0414695240745896e-05,
-      "loss": 0.0309,
+      "loss": 0.1279,
       "step": 5166
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0412839781055757e-05,
-      "loss": 0.0313,
+      "loss": 0.0203,
       "step": 5167
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0410984321365619e-05,
-      "loss": 0.0307,
+      "loss": 0.0197,
       "step": 5168
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0409128861675482e-05,
-      "loss": 0.0864,
+      "loss": 0.0947,
       "step": 5169
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0407273401985342e-05,
-      "loss": 0.1925,
+      "loss": 0.2444,
       "step": 5170
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0405417942295204e-05,
-      "loss": 0.1363,
+      "loss": 0.1115,
       "step": 5171
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0403562482605066e-05,
-      "loss": 0.0807,
+      "loss": 0.0854,
       "step": 5172
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0401707022914928e-05,
-      "loss": 0.1392,
+      "loss": 0.057,
       "step": 5173
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.039985156322479e-05,
-      "loss": 0.2575,
+      "loss": 0.0679,
       "step": 5174
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0397996103534652e-05,
-      "loss": 0.0852,
+      "loss": 0.0491,
       "step": 5175
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0396140643844512e-05,
-      "loss": 0.1419,
+      "loss": 0.0946,
       "step": 5176
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0394285184154375e-05,
-      "loss": 0.136,
+      "loss": 0.061,
       "step": 5177
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0392429724464237e-05,
-      "loss": 0.0812,
+      "loss": 0.1222,
       "step": 5178
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0390574264774097e-05,
-      "loss": 0.1972,
+      "loss": 0.0559,
       "step": 5179
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.038871880508396e-05,
-      "loss": 0.0847,
+      "loss": 0.0145,
       "step": 5180
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0386863345393823e-05,
-      "loss": 0.2479,
+      "loss": 0.0841,
       "step": 5181
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0385007885703685e-05,
-      "loss": 0.028,
+      "loss": 0.0801,
       "step": 5182
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0383152426013545e-05,
-      "loss": 0.1341,
+      "loss": 0.0768,
       "step": 5183
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0381296966323407e-05,
-      "loss": 0.0284,
+      "loss": 0.0529,
       "step": 5184
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.037944150663327e-05,
-      "loss": 0.247,
+      "loss": 0.1092,
       "step": 5185
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.037758604694313e-05,
-      "loss": 0.0816,
+      "loss": 0.0619,
       "step": 5186
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0375730587252992e-05,
-      "loss": 0.0862,
+      "loss": 0.0179,
       "step": 5187
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0373875127562854e-05,
-      "loss": 0.2472,
+      "loss": 0.1121,
       "step": 5188
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0372019667872718e-05,
-      "loss": 0.1976,
+      "loss": 0.0627,
       "step": 5189
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.0370164208182578e-05,
-      "loss": 0.0851,
+      "loss": 0.0578,
       "step": 5190
     },
     {
       "epoch": 1.44,
       "learning_rate": 1.036830874849244e-05,
-      "loss": 0.0845,
+      "loss": 0.0161,
       "step": 5191
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.03664532888023e-05,
-      "loss": 0.1895,
+      "loss": 0.0864,
       "step": 5192
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0364597829112164e-05,
-      "loss": 0.1945,
+      "loss": 0.1154,
       "step": 5193
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0362742369422025e-05,
-      "loss": 0.3612,
+      "loss": 0.1957,
       "step": 5194
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0360886909731887e-05,
-      "loss": 0.1942,
+      "loss": 0.0692,
       "step": 5195
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0359031450041747e-05,
-      "loss": 0.1408,
+      "loss": 0.0328,
       "step": 5196
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0357175990351611e-05,
-      "loss": 0.139,
+      "loss": 0.02,
       "step": 5197
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0355320530661473e-05,
-      "loss": 0.0297,
+      "loss": 0.0192,
       "step": 5198
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0353465070971333e-05,
-      "loss": 0.2994,
+      "loss": 0.0125,
       "step": 5199
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0351609611281195e-05,
-      "loss": 0.1926,
+      "loss": 0.0705,
       "step": 5200
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0349754151591059e-05,
-      "loss": 0.2515,
+      "loss": 0.0195,
       "step": 5201
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0347898691900919e-05,
-      "loss": 0.0321,
+      "loss": 0.0142,
       "step": 5202
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.034604323221078e-05,
-      "loss": 0.2439,
+      "loss": 0.0949,
       "step": 5203
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0344187772520642e-05,
-      "loss": 0.1385,
+      "loss": 0.0497,
       "step": 5204
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0342332312830506e-05,
-      "loss": 0.1325,
+      "loss": 0.0416,
       "step": 5205
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0340476853140366e-05,
-      "loss": 0.0835,
+      "loss": 0.0176,
       "step": 5206
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0338621393450228e-05,
-      "loss": 0.0879,
+      "loss": 0.1155,
       "step": 5207
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0336765933760088e-05,
-      "loss": 0.1905,
+      "loss": 0.1625,
       "step": 5208
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0334910474069952e-05,
-      "loss": 0.0894,
+      "loss": 0.1447,
       "step": 5209
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0333055014379814e-05,
-      "loss": 0.1367,
+      "loss": 0.0738,
       "step": 5210
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0331199554689676e-05,
-      "loss": 0.1885,
+      "loss": 0.0296,
       "step": 5211
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0329344094999536e-05,
-      "loss": 0.0876,
+      "loss": 0.0898,
       "step": 5212
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.03274886353094e-05,
-      "loss": 0.0898,
+      "loss": 0.0105,
       "step": 5213
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0325633175619261e-05,
-      "loss": 0.1922,
+      "loss": 0.0736,
       "step": 5214
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0323777715929121e-05,
-      "loss": 0.1358,
+      "loss": 0.0526,
       "step": 5215
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0321922256238983e-05,
-      "loss": 0.1861,
+      "loss": 0.0428,
       "step": 5216
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0320066796548847e-05,
-      "loss": 0.0878,
+      "loss": 0.0893,
       "step": 5217
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0318211336858707e-05,
-      "loss": 0.1427,
+      "loss": 0.0379,
       "step": 5218
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0316355877168569e-05,
-      "loss": 0.1907,
+      "loss": 0.0666,
       "step": 5219
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.031450041747843e-05,
-      "loss": 0.351,
+      "loss": 0.1617,
       "step": 5220
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0312644957788294e-05,
-      "loss": 0.0912,
+      "loss": 0.0586,
       "step": 5221
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0310789498098154e-05,
-      "loss": 0.2411,
+      "loss": 0.1248,
       "step": 5222
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0308934038408016e-05,
-      "loss": 0.1415,
+      "loss": 0.2083,
       "step": 5223
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0307078578717876e-05,
-      "loss": 0.1406,
+      "loss": 0.1799,
       "step": 5224
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.030522311902774e-05,
-      "loss": 0.1952,
+      "loss": 0.0257,
       "step": 5225
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0303367659337602e-05,
-      "loss": 0.0873,
+      "loss": 0.1355,
       "step": 5226
     },
     {
       "epoch": 1.45,
       "learning_rate": 1.0301512199647464e-05,
-      "loss": 0.3462,
+      "loss": 0.0827,
       "step": 5227
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0299656739957324e-05,
-      "loss": 0.2441,
+      "loss": 0.0615,
       "step": 5228
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0297801280267188e-05,
-      "loss": 0.0906,
+      "loss": 0.0227,
       "step": 5229
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.029594582057705e-05,
-      "loss": 0.1856,
+      "loss": 0.064,
       "step": 5230
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.029409036088691e-05,
-      "loss": 0.0895,
+      "loss": 0.0691,
       "step": 5231
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0292234901196771e-05,
-      "loss": 0.0855,
+      "loss": 0.0773,
       "step": 5232
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0290379441506635e-05,
-      "loss": 0.2955,
+      "loss": 0.1108,
       "step": 5233
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0288523981816497e-05,
-      "loss": 0.1867,
+      "loss": 0.0738,
       "step": 5234
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0286668522126357e-05,
-      "loss": 0.0888,
+      "loss": 0.1371,
       "step": 5235
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0284813062436219e-05,
-      "loss": 0.1398,
+      "loss": 0.0911,
       "step": 5236
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0282957602746082e-05,
-      "loss": 0.1405,
+      "loss": 0.0234,
       "step": 5237
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0281102143055943e-05,
-      "loss": 0.1895,
+      "loss": 0.1253,
       "step": 5238
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0279246683365805e-05,
-      "loss": 0.2464,
+      "loss": 0.0937,
       "step": 5239
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0277391223675666e-05,
-      "loss": 0.0912,
+      "loss": 0.1208,
       "step": 5240
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0275535763985528e-05,
-      "loss": 0.1389,
+      "loss": 0.1043,
       "step": 5241
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.027368030429539e-05,
-      "loss": 0.1365,
+      "loss": 0.0337,
       "step": 5242
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0271824844605252e-05,
-      "loss": 0.141,
+      "loss": 0.0716,
       "step": 5243
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0269969384915112e-05,
-      "loss": 0.0903,
+      "loss": 0.0189,
       "step": 5244
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0268113925224976e-05,
-      "loss": 0.091,
+      "loss": 0.0255,
       "step": 5245
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0266258465534838e-05,
-      "loss": 0.0405,
+      "loss": 0.0873,
       "step": 5246
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0264403005844698e-05,
-      "loss": 0.0391,
+      "loss": 0.0462,
       "step": 5247
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.026254754615456e-05,
-      "loss": 0.2945,
+      "loss": 0.2434,
       "step": 5248
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0260692086464423e-05,
-      "loss": 0.0917,
+      "loss": 0.122,
       "step": 5249
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0258836626774285e-05,
-      "loss": 0.243,
+      "loss": 0.0193,
       "step": 5250
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0256981167084145e-05,
-      "loss": 0.088,
+      "loss": 0.0535,
       "step": 5251
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0255125707394007e-05,
-      "loss": 0.2421,
+      "loss": 0.0492,
       "step": 5252
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.025327024770387e-05,
-      "loss": 0.1377,
+      "loss": 0.1234,
       "step": 5253
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0251414788013731e-05,
-      "loss": 0.0908,
+      "loss": 0.1003,
       "step": 5254
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0249559328323593e-05,
-      "loss": 0.0875,
+      "loss": 0.0494,
       "step": 5255
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0247703868633455e-05,
-      "loss": 0.0358,
+      "loss": 0.0838,
       "step": 5256
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0245848408943318e-05,
-      "loss": 0.1374,
+      "loss": 0.1495,
       "step": 5257
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0243992949253178e-05,
-      "loss": 0.2438,
+      "loss": 0.0631,
       "step": 5258
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.024213748956304e-05,
-      "loss": 0.1389,
+      "loss": 0.0224,
       "step": 5259
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.02402820298729e-05,
-      "loss": 0.084,
+      "loss": 0.1822,
       "step": 5260
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0238426570182764e-05,
-      "loss": 0.2988,
+      "loss": 0.0956,
       "step": 5261
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0236571110492626e-05,
-      "loss": 0.0881,
+      "loss": 0.0601,
       "step": 5262
     },
     {
       "epoch": 1.46,
       "learning_rate": 1.0234715650802488e-05,
-      "loss": 0.1384,
+      "loss": 0.0755,
       "step": 5263
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0232860191112348e-05,
-      "loss": 0.1886,
+      "loss": 0.0338,
       "step": 5264
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0231004731422211e-05,
-      "loss": 0.0884,
+      "loss": 0.0957,
       "step": 5265
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0229149271732073e-05,
-      "loss": 0.1929,
+      "loss": 0.062,
       "step": 5266
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0227293812041934e-05,
-      "loss": 0.1414,
+      "loss": 0.0506,
       "step": 5267
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0225438352351795e-05,
-      "loss": 0.1403,
+      "loss": 0.1321,
       "step": 5268
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0223582892661659e-05,
-      "loss": 0.0865,
+      "loss": 0.0213,
       "step": 5269
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0221727432971519e-05,
-      "loss": 0.1923,
+      "loss": 0.1063,
       "step": 5270
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0219871973281381e-05,
-      "loss": 0.1923,
+      "loss": 0.1454,
       "step": 5271
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0218016513591243e-05,
-      "loss": 0.139,
+      "loss": 0.1209,
       "step": 5272
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0216161053901106e-05,
-      "loss": 0.0325,
+      "loss": 0.2124,
       "step": 5273
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0214305594210967e-05,
-      "loss": 0.086,
+      "loss": 0.0187,
       "step": 5274
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0212450134520828e-05,
-      "loss": 0.088,
+      "loss": 0.2011,
       "step": 5275
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0210594674830689e-05,
-      "loss": 0.0849,
+      "loss": 0.1465,
       "step": 5276
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0208739215140552e-05,
-      "loss": 0.2488,
+      "loss": 0.114,
       "step": 5277
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0206883755450414e-05,
-      "loss": 0.238,
+      "loss": 0.0433,
       "step": 5278
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0205028295760276e-05,
-      "loss": 0.0831,
+      "loss": 0.024,
       "step": 5279
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0203172836070136e-05,
-      "loss": 0.141,
+      "loss": 0.0598,
       "step": 5280
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.020131737638e-05,
-      "loss": 0.0869,
+      "loss": 0.048,
       "step": 5281
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0199461916689862e-05,
-      "loss": 0.0847,
+      "loss": 0.042,
       "step": 5282
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0197606456999722e-05,
-      "loss": 0.086,
+      "loss": 0.0164,
       "step": 5283
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0195750997309584e-05,
-      "loss": 0.0296,
+      "loss": 0.0557,
       "step": 5284
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0193895537619447e-05,
-      "loss": 0.1929,
+      "loss": 0.0234,
       "step": 5285
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0192040077929309e-05,
-      "loss": 0.0847,
+      "loss": 0.1214,
       "step": 5286
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.019018461823917e-05,
-      "loss": 0.083,
+      "loss": 0.0605,
       "step": 5287
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0188329158549031e-05,
-      "loss": 0.0297,
+      "loss": 0.0789,
       "step": 5288
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0186473698858895e-05,
-      "loss": 0.1926,
+      "loss": 0.0201,
       "step": 5289
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0184618239168755e-05,
-      "loss": 0.0851,
+      "loss": 0.0544,
       "step": 5290
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0182762779478617e-05,
-      "loss": 0.1384,
+      "loss": 0.0243,
       "step": 5291
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0180907319788479e-05,
-      "loss": 0.1367,
+      "loss": 0.066,
       "step": 5292
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.017905186009834e-05,
-      "loss": 0.2527,
+      "loss": 0.1275,
       "step": 5293
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0177196400408202e-05,
-      "loss": 0.0835,
+      "loss": 0.0717,
       "step": 5294
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0175340940718064e-05,
-      "loss": 0.1947,
+      "loss": 0.0225,
       "step": 5295
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0173485481027924e-05,
-      "loss": 0.3699,
+      "loss": 0.2005,
       "step": 5296
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.0171630021337788e-05,
-      "loss": 0.3055,
+      "loss": 0.2521,
       "step": 5297
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.016977456164765e-05,
-      "loss": 0.141,
+      "loss": 0.0217,
       "step": 5298
     },
     {
       "epoch": 1.47,
       "learning_rate": 1.016791910195751e-05,
-      "loss": 0.3015,
+      "loss": 0.0965,
       "step": 5299
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0166063642267372e-05,
-      "loss": 0.1401,
+      "loss": 0.1854,
       "step": 5300
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0164208182577235e-05,
-      "loss": 0.0861,
+      "loss": 0.0133,
       "step": 5301
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0162352722887097e-05,
-      "loss": 0.0865,
+      "loss": 0.1378,
       "step": 5302
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0160497263196957e-05,
-      "loss": 0.0847,
+      "loss": 0.093,
       "step": 5303
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.015864180350682e-05,
-      "loss": 0.2504,
+      "loss": 0.0798,
       "step": 5304
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0156786343816683e-05,
-      "loss": 0.1422,
+      "loss": 0.0598,
       "step": 5305
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0154930884126543e-05,
-      "loss": 0.1356,
+      "loss": 0.0652,
       "step": 5306
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0153075424436405e-05,
-      "loss": 0.0877,
+      "loss": 0.0195,
       "step": 5307
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0151219964746267e-05,
-      "loss": 0.2467,
+      "loss": 0.0595,
       "step": 5308
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.014936450505613e-05,
-      "loss": 0.1417,
+      "loss": 0.0536,
       "step": 5309
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.014750904536599e-05,
-      "loss": 0.0344,
+      "loss": 0.1004,
       "step": 5310
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0145653585675852e-05,
-      "loss": 0.035,
+      "loss": 0.063,
       "step": 5311
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0143798125985713e-05,
-      "loss": 0.1904,
+      "loss": 0.0546,
       "step": 5312
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0141942666295576e-05,
-      "loss": 0.2441,
+      "loss": 0.1468,
       "step": 5313
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0140087206605438e-05,
-      "loss": 0.0852,
+      "loss": 0.0844,
       "step": 5314
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.01382317469153e-05,
-      "loss": 0.1374,
+      "loss": 0.0222,
       "step": 5315
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.013637628722516e-05,
-      "loss": 0.1383,
+      "loss": 0.0205,
       "step": 5316
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0134520827535022e-05,
-      "loss": 0.0352,
+      "loss": 0.0199,
       "step": 5317
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0132665367844886e-05,
-      "loss": 0.1374,
+      "loss": 0.1501,
       "step": 5318
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0130809908154746e-05,
-      "loss": 0.138,
+      "loss": 0.0142,
       "step": 5319
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0128954448464608e-05,
-      "loss": 0.0335,
+      "loss": 0.0837,
       "step": 5320
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.012709898877447e-05,
-      "loss": 0.0888,
+      "loss": 0.0623,
       "step": 5321
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0125243529084331e-05,
-      "loss": 0.0881,
+      "loss": 0.1017,
       "step": 5322
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0123388069394193e-05,
-      "loss": 0.1938,
+      "loss": 0.0232,
       "step": 5323
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0121532609704055e-05,
-      "loss": 0.0872,
+      "loss": 0.0243,
       "step": 5324
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0119677150013915e-05,
-      "loss": 0.1441,
+      "loss": 0.0287,
       "step": 5325
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0117821690323779e-05,
-      "loss": 0.2477,
+      "loss": 0.0475,
       "step": 5326
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.011596623063364e-05,
-      "loss": 0.1368,
+      "loss": 0.0873,
       "step": 5327
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0114110770943501e-05,
-      "loss": 0.0312,
+      "loss": 0.0875,
       "step": 5328
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0112255311253363e-05,
-      "loss": 0.0308,
+      "loss": 0.0612,
       "step": 5329
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0110399851563226e-05,
-      "loss": 0.086,
+      "loss": 0.1316,
       "step": 5330
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0108544391873088e-05,
-      "loss": 0.0846,
+      "loss": 0.016,
       "step": 5331
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0106688932182948e-05,
-      "loss": 0.1365,
+      "loss": 0.0558,
       "step": 5332
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.010483347249281e-05,
-      "loss": 0.0852,
+      "loss": 0.0134,
       "step": 5333
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0102978012802674e-05,
-      "loss": 0.135,
+      "loss": 0.0456,
       "step": 5334
     },
     {
       "epoch": 1.48,
       "learning_rate": 1.0101122553112534e-05,
-      "loss": 0.1936,
+      "loss": 0.0485,
       "step": 5335
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0099267093422396e-05,
-      "loss": 0.1939,
+      "loss": 0.0938,
       "step": 5336
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0097411633732258e-05,
-      "loss": 0.192,
+      "loss": 0.105,
       "step": 5337
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.009555617404212e-05,
-      "loss": 0.2492,
+      "loss": 0.0489,
       "step": 5338
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0093700714351981e-05,
-      "loss": 0.2516,
+      "loss": 0.0741,
       "step": 5339
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0091845254661843e-05,
-      "loss": 0.2978,
+      "loss": 0.0272,
       "step": 5340
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0089989794971703e-05,
-      "loss": 0.252,
+      "loss": 0.0963,
       "step": 5341
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0088134335281567e-05,
-      "loss": 0.1934,
+      "loss": 0.0786,
       "step": 5342
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0086278875591429e-05,
-      "loss": 0.2485,
+      "loss": 0.0283,
       "step": 5343
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0084423415901289e-05,
-      "loss": 0.0301,
+      "loss": 0.0129,
       "step": 5344
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0082567956211151e-05,
-      "loss": 0.0311,
+      "loss": 0.0183,
       "step": 5345
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0080712496521015e-05,
-      "loss": 0.1405,
+      "loss": 0.0131,
       "step": 5346
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0078857036830876e-05,
-      "loss": 0.1414,
+      "loss": 0.1178,
       "step": 5347
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0077001577140737e-05,
-      "loss": 0.1424,
+      "loss": 0.0256,
       "step": 5348
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0075146117450598e-05,
-      "loss": 0.0339,
+      "loss": 0.0316,
       "step": 5349
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0073290657760462e-05,
-      "loss": 0.1381,
+      "loss": 0.1281,
       "step": 5350
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0071435198070322e-05,
-      "loss": 0.2455,
+      "loss": 0.0367,
       "step": 5351
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0069579738380184e-05,
-      "loss": 0.1379,
+      "loss": 0.0323,
       "step": 5352
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0067724278690046e-05,
-      "loss": 0.3518,
+      "loss": 0.0146,
       "step": 5353
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.006586881899991e-05,
-      "loss": 0.088,
+      "loss": 0.0533,
       "step": 5354
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.006401335930977e-05,
-      "loss": 0.1913,
+      "loss": 0.4008,
       "step": 5355
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0062157899619632e-05,
-      "loss": 0.134,
+      "loss": 0.0963,
       "step": 5356
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0060302439929492e-05,
-      "loss": 0.1918,
+      "loss": 0.1192,
       "step": 5357
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0058446980239355e-05,
-      "loss": 0.0879,
+      "loss": 0.1775,
       "step": 5358
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0056591520549217e-05,
-      "loss": 0.1904,
+      "loss": 0.0556,
       "step": 5359
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0054736060859079e-05,
-      "loss": 0.2474,
+      "loss": 0.1849,
       "step": 5360
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.005288060116894e-05,
-      "loss": 0.1944,
+      "loss": 0.1592,
       "step": 5361
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0051025141478803e-05,
-      "loss": 0.1431,
+      "loss": 0.0896,
       "step": 5362
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0049169681788665e-05,
-      "loss": 0.09,
+      "loss": 0.1752,
       "step": 5363
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0047314222098525e-05,
-      "loss": 0.1929,
+      "loss": 0.1845,
       "step": 5364
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0045458762408387e-05,
-      "loss": 0.0873,
+      "loss": 0.1078,
       "step": 5365
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.004360330271825e-05,
-      "loss": 0.1416,
+      "loss": 0.0294,
       "step": 5366
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.004174784302811e-05,
-      "loss": 0.1394,
+      "loss": 0.0365,
       "step": 5367
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0039892383337972e-05,
-      "loss": 0.2438,
+      "loss": 0.0634,
       "step": 5368
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0038036923647834e-05,
-      "loss": 0.1868,
+      "loss": 0.0344,
       "step": 5369
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0036181463957698e-05,
-      "loss": 0.1381,
+      "loss": 0.142,
       "step": 5370
     },
     {
       "epoch": 1.49,
       "learning_rate": 1.0034326004267558e-05,
-      "loss": 0.196,
+      "loss": 0.0961,
       "step": 5371
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.003247054457742e-05,
-      "loss": 0.146,
+      "loss": 0.0137,
       "step": 5372
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.003061508488728e-05,
-      "loss": 0.0882,
+      "loss": 0.1149,
       "step": 5373
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0028759625197144e-05,
-      "loss": 0.1398,
+      "loss": 0.0805,
       "step": 5374
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0026904165507005e-05,
-      "loss": 0.1429,
+      "loss": 0.0738,
       "step": 5375
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0025048705816867e-05,
-      "loss": 0.1422,
+      "loss": 0.0667,
       "step": 5376
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0023193246126727e-05,
-      "loss": 0.2388,
+      "loss": 0.1225,
       "step": 5377
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0021337786436591e-05,
-      "loss": 0.1933,
+      "loss": 0.0315,
       "step": 5378
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0019482326746453e-05,
-      "loss": 0.039,
+      "loss": 0.0452,
       "step": 5379
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0017626867056313e-05,
-      "loss": 0.191,
+      "loss": 0.027,
       "step": 5380
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0015771407366175e-05,
-      "loss": 0.1406,
+      "loss": 0.1206,
       "step": 5381
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0013915947676038e-05,
-      "loss": 0.1401,
+      "loss": 0.1174,
       "step": 5382
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.00120604879859e-05,
-      "loss": 0.0913,
+      "loss": 0.1267,
       "step": 5383
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.001020502829576e-05,
-      "loss": 0.038,
+      "loss": 0.0457,
       "step": 5384
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0008349568605622e-05,
-      "loss": 0.0877,
+      "loss": 0.0605,
       "step": 5385
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0006494108915486e-05,
-      "loss": 0.0371,
+      "loss": 0.0175,
       "step": 5386
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0004638649225346e-05,
-      "loss": 0.1911,
+      "loss": 0.0265,
       "step": 5387
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.0002783189535208e-05,
-      "loss": 0.0885,
+      "loss": 0.1394,
       "step": 5388
     },
     {
       "epoch": 1.5,
       "learning_rate": 1.000092772984507e-05,
-      "loss": 0.1943,
+      "loss": 0.1051,
       "step": 5389
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.999072270154932e-06,
-      "loss": 0.2394,
+      "loss": 0.0658,
       "step": 5390
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.997216810464794e-06,
-      "loss": 0.0355,
+      "loss": 0.0871,
       "step": 5391
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.995361350774655e-06,
-      "loss": 0.0347,
+      "loss": 0.1016,
       "step": 5392
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.993505891084517e-06,
-      "loss": 0.033,
+      "loss": 0.1068,
       "step": 5393
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.99165043139438e-06,
-      "loss": 0.1998,
+      "loss": 0.0232,
       "step": 5394
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.989794971704241e-06,
-      "loss": 0.0325,
+      "loss": 0.1059,
       "step": 5395
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.987939512014101e-06,
-      "loss": 0.0861,
+      "loss": 0.0237,
       "step": 5396
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.986084052323965e-06,
-      "loss": 0.0292,
+      "loss": 0.0523,
       "step": 5397
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.984228592633825e-06,
-      "loss": 0.1958,
+      "loss": 0.1995,
       "step": 5398
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.982373132943689e-06,
-      "loss": 0.1935,
+      "loss": 0.2027,
       "step": 5399
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.980517673253549e-06,
-      "loss": 0.085,
+      "loss": 0.0592,
       "step": 5400
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.97866221356341e-06,
-      "loss": 0.3077,
+      "loss": 0.0788,
       "step": 5401
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.976806753873273e-06,
-      "loss": 0.0816,
+      "loss": 0.0636,
       "step": 5402
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.974951294183134e-06,
-      "loss": 0.1981,
+      "loss": 0.1771,
       "step": 5403
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.973095834492996e-06,
-      "loss": 0.0285,
+      "loss": 0.0252,
       "step": 5404
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.971240374802858e-06,
-      "loss": 0.4264,
+      "loss": 0.0422,
       "step": 5405
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.96938491511272e-06,
-      "loss": 0.1404,
+      "loss": 0.0228,
       "step": 5406
     },
     {
       "epoch": 1.5,
       "learning_rate": 9.967529455422582e-06,
-      "loss": 0.2994,
+      "loss": 0.1457,
       "step": 5407
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.965673995732444e-06,
-      "loss": 0.0833,
+      "loss": 0.1166,
       "step": 5408
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.963818536042306e-06,
-      "loss": 0.0833,
+      "loss": 0.0245,
       "step": 5409
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.961963076352167e-06,
-      "loss": 0.0864,
+      "loss": 0.0933,
       "step": 5410
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.96010761666203e-06,
-      "loss": 0.0833,
+      "loss": 0.0106,
       "step": 5411
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.958252156971891e-06,
-      "loss": 0.1389,
+      "loss": 0.0207,
       "step": 5412
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.956396697281753e-06,
-      "loss": 0.1384,
+      "loss": 0.0906,
       "step": 5413
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.954541237591613e-06,
-      "loss": 0.1425,
+      "loss": 0.0473,
       "step": 5414
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.952685777901477e-06,
-      "loss": 0.1389,
+      "loss": 0.0796,
       "step": 5415
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.950830318211337e-06,
-      "loss": 0.028,
+      "loss": 0.086,
       "step": 5416
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.9489748585212e-06,
-      "loss": 0.1991,
+      "loss": 0.0634,
       "step": 5417
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.94711939883106e-06,
-      "loss": 0.3128,
+      "loss": 0.0817,
       "step": 5418
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.945263939140923e-06,
-      "loss": 0.0293,
+      "loss": 0.1864,
       "step": 5419
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.943408479450784e-06,
-      "loss": 0.1404,
+      "loss": 0.0927,
       "step": 5420
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.941553019760646e-06,
-      "loss": 0.0833,
+      "loss": 0.1552,
       "step": 5421
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.939697560070508e-06,
-      "loss": 0.0855,
+      "loss": 0.0468,
       "step": 5422
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.93784210038037e-06,
-      "loss": 0.1938,
+      "loss": 0.0231,
       "step": 5423
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.935986640690232e-06,
-      "loss": 0.0857,
+      "loss": 0.0296,
       "step": 5424
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.934131181000094e-06,
-      "loss": 0.1372,
+      "loss": 0.0765,
       "step": 5425
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.932275721309956e-06,
-      "loss": 0.1408,
+      "loss": 0.0147,
       "step": 5426
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.930420261619818e-06,
-      "loss": 0.1953,
+      "loss": 0.0758,
       "step": 5427
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.92856480192968e-06,
-      "loss": 0.0856,
+      "loss": 0.0134,
       "step": 5428
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.926709342239541e-06,
-      "loss": 0.1423,
+      "loss": 0.0675,
       "step": 5429
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.924853882549401e-06,
-      "loss": 0.1419,
+      "loss": 0.102,
       "step": 5430
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.922998422859265e-06,
-      "loss": 0.1425,
+      "loss": 0.1058,
       "step": 5431
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.921142963169125e-06,
-      "loss": 0.1945,
+      "loss": 0.1405,
       "step": 5432
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.919287503478989e-06,
-      "loss": 0.1977,
+      "loss": 0.0552,
       "step": 5433
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.917432043788849e-06,
-      "loss": 0.1365,
+      "loss": 0.0958,
       "step": 5434
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.915576584098713e-06,
-      "loss": 0.0841,
+      "loss": 0.0263,
       "step": 5435
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.913721124408573e-06,
-      "loss": 0.087,
+      "loss": 0.1458,
       "step": 5436
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.911865664718435e-06,
-      "loss": 0.1416,
+      "loss": 0.0892,
       "step": 5437
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.910010205028296e-06,
-      "loss": 0.1379,
+      "loss": 0.0681,
       "step": 5438
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.908154745338158e-06,
-      "loss": 0.1979,
+      "loss": 0.0418,
       "step": 5439
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.90629928564802e-06,
-      "loss": 0.086,
+      "loss": 0.0164,
       "step": 5440
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.904443825957882e-06,
-      "loss": 0.0845,
+      "loss": 0.0339,
       "step": 5441
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.902588366267744e-06,
-      "loss": 0.1895,
+      "loss": 0.0513,
       "step": 5442
     },
     {
       "epoch": 1.51,
       "learning_rate": 9.900732906577606e-06,
-      "loss": 0.1357,
+      "loss": 0.0221,
       "step": 5443
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.898877446887468e-06,
-      "loss": 0.1396,
+      "loss": 0.0917,
       "step": 5444
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.89702198719733e-06,
-      "loss": 0.0862,
+      "loss": 0.0922,
       "step": 5445
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.895166527507191e-06,
-      "loss": 0.1402,
+      "loss": 0.0625,
       "step": 5446
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.893311067817053e-06,
-      "loss": 0.0292,
+      "loss": 0.0411,
       "step": 5447
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.891455608126913e-06,
-      "loss": 0.0848,
+      "loss": 0.0487,
       "step": 5448
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.889600148436777e-06,
-      "loss": 0.0837,
+      "loss": 0.0155,
       "step": 5449
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.887744688746637e-06,
-      "loss": 0.0848,
+      "loss": 0.1506,
       "step": 5450
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.8858892290565e-06,
-      "loss": 0.0859,
+      "loss": 0.056,
       "step": 5451
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.884033769366361e-06,
-      "loss": 0.1417,
+      "loss": 0.008,
       "step": 5452
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.882178309676223e-06,
-      "loss": 0.1388,
+      "loss": 0.0938,
       "step": 5453
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.880322849986085e-06,
-      "loss": 0.2487,
+      "loss": 0.0673,
       "step": 5454
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.878467390295947e-06,
-      "loss": 0.1927,
+      "loss": 0.0628,
       "step": 5455
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.876611930605808e-06,
-      "loss": 0.1983,
+      "loss": 0.0094,
       "step": 5456
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.87475647091567e-06,
-      "loss": 0.0851,
+      "loss": 0.0548,
       "step": 5457
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.872901011225532e-06,
-      "loss": 0.0292,
+      "loss": 0.211,
       "step": 5458
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.871045551535394e-06,
-      "loss": 0.0292,
+      "loss": 0.1662,
       "step": 5459
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.869190091845256e-06,
-      "loss": 0.0289,
+      "loss": 0.0568,
       "step": 5460
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.867334632155118e-06,
-      "loss": 0.1406,
+      "loss": 0.115,
       "step": 5461
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.86547917246498e-06,
-      "loss": 0.1881,
+      "loss": 0.0145,
       "step": 5462
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.863623712774842e-06,
-      "loss": 0.0835,
+      "loss": 0.2141,
       "step": 5463
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.861768253084702e-06,
-      "loss": 0.1943,
+      "loss": 0.0652,
       "step": 5464
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.859912793394565e-06,
-      "loss": 0.1946,
+      "loss": 0.0179,
       "step": 5465
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.858057333704425e-06,
-      "loss": 0.1402,
+      "loss": 0.0558,
       "step": 5466
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.856201874014287e-06,
-      "loss": 0.0826,
+      "loss": 0.0212,
       "step": 5467
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.85434641432415e-06,
-      "loss": 0.0814,
+      "loss": 0.0876,
       "step": 5468
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.852490954634011e-06,
-      "loss": 0.1391,
+      "loss": 0.0474,
       "step": 5469
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.850635494943873e-06,
-      "loss": 0.2504,
+      "loss": 0.0605,
       "step": 5470
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.848780035253735e-06,
-      "loss": 0.3615,
+      "loss": 0.0139,
       "step": 5471
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.846924575563597e-06,
-      "loss": 0.0814,
+      "loss": 0.0652,
       "step": 5472
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.845069115873459e-06,
-      "loss": 0.4231,
+      "loss": 0.1315,
       "step": 5473
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.84321365618332e-06,
-      "loss": 0.1422,
+      "loss": 0.1499,
       "step": 5474
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.841358196493182e-06,
-      "loss": 0.1396,
+      "loss": 0.0144,
       "step": 5475
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.839502736803044e-06,
-      "loss": 0.0853,
+      "loss": 0.1521,
       "step": 5476
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.837647277112904e-06,
-      "loss": 0.1365,
+      "loss": 0.127,
       "step": 5477
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.835791817422768e-06,
-      "loss": 0.1938,
+      "loss": 0.0159,
       "step": 5478
     },
     {
       "epoch": 1.52,
       "learning_rate": 9.833936357732628e-06,
-      "loss": 0.033,
+      "loss": 0.0279,
       "step": 5479
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.832080898042492e-06,
-      "loss": 0.1925,
+      "loss": 0.0646,
       "step": 5480
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.830225438352352e-06,
-      "loss": 0.2465,
+      "loss": 0.0579,
       "step": 5481
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.828369978662214e-06,
-      "loss": 0.2423,
+      "loss": 0.0177,
       "step": 5482
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.826514518972076e-06,
-      "loss": 0.0853,
+      "loss": 0.0964,
       "step": 5483
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.824659059281937e-06,
-      "loss": 0.0876,
+      "loss": 0.0156,
       "step": 5484
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.8228035995918e-06,
-      "loss": 0.1429,
+      "loss": 0.0171,
       "step": 5485
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.820948139901661e-06,
-      "loss": 0.0844,
+      "loss": 0.0488,
       "step": 5486
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.819092680211523e-06,
-      "loss": 0.084,
+      "loss": 0.045,
       "step": 5487
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.817237220521385e-06,
-      "loss": 0.3522,
+      "loss": 0.0122,
       "step": 5488
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.815381760831247e-06,
-      "loss": 0.0857,
+      "loss": 0.076,
       "step": 5489
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.813526301141109e-06,
-      "loss": 0.1414,
+      "loss": 0.0625,
       "step": 5490
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.81167084145097e-06,
-      "loss": 0.1434,
+      "loss": 0.1299,
       "step": 5491
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.809815381760832e-06,
-      "loss": 0.0896,
+      "loss": 0.058,
       "step": 5492
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.807959922070693e-06,
-      "loss": 0.187,
+      "loss": 0.0938,
       "step": 5493
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.806104462380556e-06,
-      "loss": 0.1397,
+      "loss": 0.0473,
       "step": 5494
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.804249002690416e-06,
-      "loss": 0.1402,
+      "loss": 0.1067,
       "step": 5495
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.80239354300028e-06,
-      "loss": 0.2429,
+      "loss": 0.0892,
       "step": 5496
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.80053808331014e-06,
-      "loss": 0.0352,
+      "loss": 0.0993,
       "step": 5497
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.798682623620004e-06,
-      "loss": 0.1421,
+      "loss": 0.0673,
       "step": 5498
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.796827163929864e-06,
-      "loss": 0.1903,
+      "loss": 0.0598,
       "step": 5499
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.794971704239726e-06,
-      "loss": 0.2899,
+      "loss": 0.0498,
       "step": 5500
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.793116244549588e-06,
-      "loss": 0.1418,
+      "loss": 0.0244,
       "step": 5501
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.79126078485945e-06,
-      "loss": 0.1406,
+      "loss": 0.0168,
       "step": 5502
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.789405325169311e-06,
-      "loss": 0.1434,
+      "loss": 0.1873,
       "step": 5503
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.787549865479173e-06,
-      "loss": 0.1396,
+      "loss": 0.0577,
       "step": 5504
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.785694405789035e-06,
-      "loss": 0.0373,
+      "loss": 0.085,
       "step": 5505
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.783838946098897e-06,
-      "loss": 0.0356,
+      "loss": 0.1775,
       "step": 5506
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.781983486408759e-06,
-      "loss": 0.1887,
+      "loss": 0.0865,
       "step": 5507
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.78012802671862e-06,
-      "loss": 0.1388,
+      "loss": 0.0533,
       "step": 5508
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.778272567028482e-06,
-      "loss": 0.0351,
+      "loss": 0.0562,
       "step": 5509
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.776417107338344e-06,
-      "loss": 0.0872,
+      "loss": 0.2182,
       "step": 5510
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.774561647648205e-06,
-      "loss": 0.0866,
+      "loss": 0.0609,
       "step": 5511
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.772706187958068e-06,
-      "loss": 0.1415,
+      "loss": 0.1132,
       "step": 5512
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.770850728267928e-06,
-      "loss": 0.144,
+      "loss": 0.0737,
       "step": 5513
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.768995268577792e-06,
-      "loss": 0.1434,
+      "loss": 0.0233,
       "step": 5514
     },
     {
       "epoch": 1.53,
       "learning_rate": 9.767139808887652e-06,
-      "loss": 0.1389,
+      "loss": 0.1068,
       "step": 5515
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.765284349197514e-06,
-      "loss": 0.1902,
+      "loss": 0.0541,
       "step": 5516
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.763428889507376e-06,
-      "loss": 0.1952,
+      "loss": 0.0295,
       "step": 5517
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.761573429817238e-06,
-      "loss": 0.0843,
+      "loss": 0.021,
       "step": 5518
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.7597179701271e-06,
-      "loss": 0.0869,
+      "loss": 0.0479,
       "step": 5519
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.757862510436961e-06,
-      "loss": 0.1929,
+      "loss": 0.1204,
       "step": 5520
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.756007050746823e-06,
-      "loss": 0.0861,
+      "loss": 0.1299,
       "step": 5521
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.754151591056685e-06,
-      "loss": 0.2458,
+      "loss": 0.1432,
       "step": 5522
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.752296131366547e-06,
-      "loss": 0.0302,
+      "loss": 0.0273,
       "step": 5523
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.750440671676409e-06,
-      "loss": 0.0294,
+      "loss": 0.0888,
       "step": 5524
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.74858521198627e-06,
-      "loss": 0.3054,
+      "loss": 0.0566,
       "step": 5525
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.746729752296133e-06,
-      "loss": 0.1346,
+      "loss": 0.1232,
       "step": 5526
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.744874292605993e-06,
-      "loss": 0.1937,
+      "loss": 0.1358,
       "step": 5527
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.743018832915856e-06,
-      "loss": 0.247,
+      "loss": 0.0975,
       "step": 5528
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.741163373225717e-06,
-      "loss": 0.134,
+      "loss": 0.0563,
       "step": 5529
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.73930791353558e-06,
-      "loss": 0.0305,
+      "loss": 0.0186,
       "step": 5530
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.73745245384544e-06,
-      "loss": 0.0848,
+      "loss": 0.0534,
       "step": 5531
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.735596994155304e-06,
-      "loss": 0.3026,
+      "loss": 0.0523,
       "step": 5532
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.733741534465164e-06,
-      "loss": 0.3013,
+      "loss": 0.0217,
       "step": 5533
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.731886074775026e-06,
-      "loss": 0.0848,
+      "loss": 0.019,
       "step": 5534
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.730030615084888e-06,
-      "loss": 0.2941,
+      "loss": 0.0146,
       "step": 5535
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.72817515539475e-06,
-      "loss": 0.084,
+      "loss": 0.0639,
       "step": 5536
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.726319695704611e-06,
-      "loss": 0.0323,
+      "loss": 0.0815,
       "step": 5537
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.724464236014473e-06,
-      "loss": 0.1907,
+      "loss": 0.1084,
       "step": 5538
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.722608776324335e-06,
-      "loss": 0.0333,
+      "loss": 0.1023,
       "step": 5539
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.720753316634197e-06,
-      "loss": 0.1443,
+      "loss": 0.0236,
       "step": 5540
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.718897856944059e-06,
-      "loss": 0.0861,
+      "loss": 0.0135,
       "step": 5541
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.71704239725392e-06,
-      "loss": 0.2477,
+      "loss": 0.0131,
       "step": 5542
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.715186937563783e-06,
-      "loss": 0.0918,
+      "loss": 0.1096,
       "step": 5543
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.713331477873645e-06,
-      "loss": 0.194,
+      "loss": 0.0346,
       "step": 5544
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.711476018183505e-06,
-      "loss": 0.0874,
+      "loss": 0.0551,
       "step": 5545
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.709620558493368e-06,
-      "loss": 0.0855,
+      "loss": 0.0481,
       "step": 5546
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.707765098803228e-06,
-      "loss": 0.2481,
+      "loss": 0.0205,
       "step": 5547
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.705909639113092e-06,
-      "loss": 0.1401,
+      "loss": 0.0426,
       "step": 5548
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.704054179422952e-06,
-      "loss": 0.2467,
+      "loss": 0.0968,
       "step": 5549
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.702198719732814e-06,
-      "loss": 0.2447,
+      "loss": 0.0401,
       "step": 5550
     },
     {
       "epoch": 1.54,
       "learning_rate": 9.700343260042676e-06,
-      "loss": 0.1369,
+      "loss": 0.0457,
       "step": 5551
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.698487800352538e-06,
-      "loss": 0.0857,
+      "loss": 0.0473,
       "step": 5552
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.6966323406624e-06,
-      "loss": 0.1947,
+      "loss": 0.0131,
       "step": 5553
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.694776880972262e-06,
-      "loss": 0.0832,
+      "loss": 0.0094,
       "step": 5554
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.692921421282123e-06,
-      "loss": 0.253,
+      "loss": 0.0505,
       "step": 5555
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.691065961591985e-06,
-      "loss": 0.1387,
+      "loss": 0.053,
       "step": 5556
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.689210501901847e-06,
-      "loss": 0.1382,
+      "loss": 0.0816,
       "step": 5557
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.687355042211709e-06,
-      "loss": 0.0358,
+      "loss": 0.0826,
       "step": 5558
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.685499582521571e-06,
-      "loss": 0.0868,
+      "loss": 0.1334,
       "step": 5559
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.683644122831433e-06,
-      "loss": 0.1947,
+      "loss": 0.1038,
       "step": 5560
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.681788663141295e-06,
-      "loss": 0.1951,
+      "loss": 0.1431,
       "step": 5561
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.679933203451157e-06,
-      "loss": 0.1409,
+      "loss": 0.0615,
       "step": 5562
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.678077743761017e-06,
-      "loss": 0.1931,
+      "loss": 0.1222,
       "step": 5563
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.67622228407088e-06,
-      "loss": 0.1894,
+      "loss": 0.0163,
       "step": 5564
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.67436682438074e-06,
-      "loss": 0.1924,
+      "loss": 0.0455,
       "step": 5565
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.672511364690604e-06,
-      "loss": 0.1374,
+      "loss": 0.0563,
       "step": 5566
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.670655905000464e-06,
-      "loss": 0.2413,
+      "loss": 0.1382,
       "step": 5567
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.668800445310326e-06,
-      "loss": 0.2416,
+      "loss": 0.0669,
       "step": 5568
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.666944985620188e-06,
-      "loss": 0.1971,
+      "loss": 0.0321,
       "step": 5569
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.66508952593005e-06,
-      "loss": 0.0891,
+      "loss": 0.0961,
       "step": 5570
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.663234066239912e-06,
-      "loss": 0.1407,
+      "loss": 0.1527,
       "step": 5571
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.661378606549774e-06,
-      "loss": 0.0919,
+      "loss": 0.124,
       "step": 5572
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.659523146859635e-06,
-      "loss": 0.0396,
+      "loss": 0.2176,
       "step": 5573
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.657667687169497e-06,
-      "loss": 0.1908,
+      "loss": 0.1166,
       "step": 5574
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.65581222747936e-06,
-      "loss": 0.1406,
+      "loss": 0.0184,
       "step": 5575
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.653956767789221e-06,
-      "loss": 0.14,
+      "loss": 0.0788,
       "step": 5576
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.652101308099083e-06,
-      "loss": 0.1378,
+      "loss": 0.0254,
       "step": 5577
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.650245848408945e-06,
-      "loss": 0.0886,
+      "loss": 0.0258,
       "step": 5578
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.648390388718805e-06,
-      "loss": 0.0898,
+      "loss": 0.2235,
       "step": 5579
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.646534929028669e-06,
-      "loss": 0.2436,
+      "loss": 0.0982,
       "step": 5580
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.644679469338529e-06,
-      "loss": 0.1932,
+      "loss": 0.0918,
       "step": 5581
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.642824009648392e-06,
-      "loss": 0.2423,
+      "loss": 0.0211,
       "step": 5582
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.640968549958252e-06,
-      "loss": 0.2424,
+      "loss": 0.0197,
       "step": 5583
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.639113090268114e-06,
-      "loss": 0.0353,
+      "loss": 0.0475,
       "step": 5584
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.637257630577976e-06,
-      "loss": 0.1968,
+      "loss": 0.0249,
       "step": 5585
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.635402170887838e-06,
-      "loss": 0.1944,
+      "loss": 0.1544,
       "step": 5586
     },
     {
       "epoch": 1.55,
       "learning_rate": 9.6335467111977e-06,
-      "loss": 0.0886,
+      "loss": 0.1587,
       "step": 5587
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.631691251507562e-06,
-      "loss": 0.1404,
+      "loss": 0.0986,
       "step": 5588
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.629835791817424e-06,
-      "loss": 0.1914,
+      "loss": 0.0515,
       "step": 5589
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.627980332127286e-06,
-      "loss": 0.1948,
+      "loss": 0.049,
       "step": 5590
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.626124872437147e-06,
-      "loss": 0.2432,
+      "loss": 0.0841,
       "step": 5591
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.62426941274701e-06,
-      "loss": 0.1425,
+      "loss": 0.0911,
       "step": 5592
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.622413953056871e-06,
-      "loss": 0.1388,
+      "loss": 0.0742,
       "step": 5593
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.620558493366733e-06,
-      "loss": 0.1418,
+      "loss": 0.0453,
       "step": 5594
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.618703033676595e-06,
-      "loss": 0.1944,
+      "loss": 0.1268,
       "step": 5595
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.616847573986457e-06,
-      "loss": 0.2491,
+      "loss": 0.0846,
       "step": 5596
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.614992114296317e-06,
-      "loss": 0.1382,
+      "loss": 0.0769,
       "step": 5597
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.61313665460618e-06,
-      "loss": 0.0371,
+      "loss": 0.0136,
       "step": 5598
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.61128119491604e-06,
-      "loss": 0.0374,
+      "loss": 0.0959,
       "step": 5599
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.609425735225904e-06,
-      "loss": 0.0901,
+      "loss": 0.0122,
       "step": 5600
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.607570275535764e-06,
-      "loss": 0.139,
+      "loss": 0.0387,
       "step": 5601
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.605714815845626e-06,
-      "loss": 0.0871,
+      "loss": 0.0453,
       "step": 5602
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.603859356155488e-06,
-      "loss": 0.0885,
+      "loss": 0.0137,
       "step": 5603
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.60200389646535e-06,
-      "loss": 0.0877,
+      "loss": 0.101,
       "step": 5604
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.600148436775212e-06,
-      "loss": 0.0349,
+      "loss": 0.0501,
       "step": 5605
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.598292977085074e-06,
-      "loss": 0.1942,
+      "loss": 0.0559,
       "step": 5606
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.596437517394936e-06,
-      "loss": 0.1363,
+      "loss": 0.0137,
       "step": 5607
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.594582057704798e-06,
-      "loss": 0.033,
+      "loss": 0.0093,
       "step": 5608
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.59272659801466e-06,
-      "loss": 0.191,
+      "loss": 0.0535,
       "step": 5609
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.590871138324521e-06,
-      "loss": 0.1428,
+      "loss": 0.0659,
       "step": 5610
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.589015678634383e-06,
-      "loss": 0.2918,
+      "loss": 0.0086,
       "step": 5611
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.587160218944245e-06,
-      "loss": 0.0861,
+      "loss": 0.0105,
       "step": 5612
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.585304759254105e-06,
-      "loss": 0.1438,
+      "loss": 0.1615,
       "step": 5613
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.583449299563969e-06,
-      "loss": 0.031,
+      "loss": 0.0125,
       "step": 5614
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.581593839873829e-06,
-      "loss": 0.0894,
+      "loss": 0.1265,
       "step": 5615
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.579738380183692e-06,
-      "loss": 0.0846,
+      "loss": 0.0565,
       "step": 5616
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.577882920493553e-06,
-      "loss": 0.0817,
+      "loss": 0.1446,
       "step": 5617
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.576027460803415e-06,
-      "loss": 0.0302,
+      "loss": 0.1354,
       "step": 5618
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.574172001113276e-06,
-      "loss": 0.1392,
+      "loss": 0.1448,
       "step": 5619
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.572316541423138e-06,
-      "loss": 0.0851,
+      "loss": 0.0678,
       "step": 5620
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.570461081733e-06,
-      "loss": 0.0864,
+      "loss": 0.09,
       "step": 5621
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.568605622042862e-06,
-      "loss": 0.1946,
+      "loss": 0.1213,
       "step": 5622
     },
     {
       "epoch": 1.56,
       "learning_rate": 9.566750162352724e-06,
-      "loss": 0.141,
+      "loss": 0.0155,
       "step": 5623
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.564894702662586e-06,
-      "loss": 0.029,
+      "loss": 0.0201,
       "step": 5624
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.563039242972448e-06,
-      "loss": 0.0803,
+      "loss": 0.058,
       "step": 5625
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.561183783282308e-06,
-      "loss": 0.2504,
+      "loss": 0.0195,
       "step": 5626
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.559328323592171e-06,
-      "loss": 0.1937,
+      "loss": 0.0125,
       "step": 5627
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.557472863902032e-06,
-      "loss": 0.139,
+      "loss": 0.0179,
       "step": 5628
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.555617404211895e-06,
-      "loss": 0.3087,
+      "loss": 0.0178,
       "step": 5629
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.553761944521755e-06,
-      "loss": 0.0842,
+      "loss": 0.0627,
       "step": 5630
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.551906484831617e-06,
-      "loss": 0.0274,
+      "loss": 0.0193,
       "step": 5631
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.550051025141479e-06,
-      "loss": 0.0825,
+      "loss": 0.0182,
       "step": 5632
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.548195565451341e-06,
-      "loss": 0.0269,
+      "loss": 0.0807,
       "step": 5633
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.546340105761203e-06,
-      "loss": 0.2467,
+      "loss": 0.1807,
       "step": 5634
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.544484646071065e-06,
-      "loss": 0.0854,
+      "loss": 0.1398,
       "step": 5635
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.542629186380927e-06,
-      "loss": 0.1329,
+      "loss": 0.012,
       "step": 5636
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.540773726690788e-06,
-      "loss": 0.1405,
+      "loss": 0.1211,
       "step": 5637
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.53891826700065e-06,
-      "loss": 0.1396,
+      "loss": 0.0614,
       "step": 5638
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.537062807310512e-06,
-      "loss": 0.1408,
+      "loss": 0.0202,
       "step": 5639
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.535207347620374e-06,
-      "loss": 0.0845,
+      "loss": 0.0193,
       "step": 5640
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.533351887930236e-06,
-      "loss": 0.0814,
+      "loss": 0.0682,
       "step": 5641
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.531496428240096e-06,
-      "loss": 0.1393,
+      "loss": 0.013,
       "step": 5642
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.52964096854996e-06,
-      "loss": 0.1398,
+      "loss": 0.0645,
       "step": 5643
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.52778550885982e-06,
-      "loss": 0.1417,
+      "loss": 0.0705,
       "step": 5644
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.525930049169683e-06,
-      "loss": 0.085,
+      "loss": 0.011,
       "step": 5645
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.524074589479544e-06,
-      "loss": 0.0855,
+      "loss": 0.1168,
       "step": 5646
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.522219129789405e-06,
-      "loss": 0.195,
+      "loss": 0.0505,
       "step": 5647
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.520363670099267e-06,
-      "loss": 0.1958,
+      "loss": 0.0664,
       "step": 5648
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.518508210409129e-06,
-      "loss": 0.1451,
+      "loss": 0.016,
       "step": 5649
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.516652750718991e-06,
-      "loss": 0.1909,
+      "loss": 0.12,
       "step": 5650
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.514797291028853e-06,
-      "loss": 0.2536,
+      "loss": 0.1022,
       "step": 5651
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.512941831338715e-06,
-      "loss": 0.0824,
+      "loss": 0.1301,
       "step": 5652
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.511086371648577e-06,
-      "loss": 0.2496,
+      "loss": 0.1377,
       "step": 5653
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.509230911958438e-06,
-      "loss": 0.0839,
+      "loss": 0.0129,
       "step": 5654
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.5073754522683e-06,
-      "loss": 0.3104,
+      "loss": 0.0092,
       "step": 5655
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.505519992578162e-06,
-      "loss": 0.1969,
+      "loss": 0.1208,
       "step": 5656
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.503664532888024e-06,
-      "loss": 0.2485,
+      "loss": 0.0207,
       "step": 5657
     },
     {
       "epoch": 1.57,
       "learning_rate": 9.501809073197886e-06,
-      "loss": 0.1972,
+      "loss": 0.0197,
       "step": 5658
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.499953613507748e-06,
-      "loss": 0.1421,
+      "loss": 0.0237,
       "step": 5659
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.498098153817608e-06,
-      "loss": 0.1895,
+      "loss": 0.0183,
       "step": 5660
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.496242694127472e-06,
-      "loss": 0.0861,
+      "loss": 0.0114,
       "step": 5661
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.494387234437332e-06,
-      "loss": 0.034,
+      "loss": 0.0594,
       "step": 5662
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.492531774747195e-06,
-      "loss": 0.2391,
+      "loss": 0.0631,
       "step": 5663
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.490676315057055e-06,
-      "loss": 0.1394,
+      "loss": 0.0766,
       "step": 5664
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.488820855366917e-06,
-      "loss": 0.1916,
+      "loss": 0.0171,
       "step": 5665
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.48696539567678e-06,
-      "loss": 0.134,
+      "loss": 0.1425,
       "step": 5666
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.485109935986641e-06,
-      "loss": 0.137,
+      "loss": 0.1025,
       "step": 5667
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.483254476296503e-06,
-      "loss": 0.0896,
+      "loss": 0.1075,
       "step": 5668
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.481399016606365e-06,
-      "loss": 0.1928,
+      "loss": 0.1141,
       "step": 5669
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.479543556916227e-06,
-      "loss": 0.1896,
+      "loss": 0.1164,
       "step": 5670
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.477688097226089e-06,
-      "loss": 0.0868,
+      "loss": 0.0929,
       "step": 5671
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.47583263753595e-06,
-      "loss": 0.1405,
+      "loss": 0.0489,
       "step": 5672
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.473977177845812e-06,
-      "loss": 0.1926,
+      "loss": 0.016,
       "step": 5673
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.472121718155674e-06,
-      "loss": 0.0857,
+      "loss": 0.0164,
       "step": 5674
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.470266258465536e-06,
-      "loss": 0.1414,
+      "loss": 0.0763,
       "step": 5675
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.468410798775396e-06,
-      "loss": 0.1881,
+      "loss": 0.0123,
       "step": 5676
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.46655533908526e-06,
-      "loss": 0.0373,
+      "loss": 0.0612,
       "step": 5677
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.46469987939512e-06,
-      "loss": 0.1928,
+      "loss": 0.0145,
       "step": 5678
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.462844419704984e-06,
-      "loss": 0.0371,
+      "loss": 0.0619,
       "step": 5679
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.460988960014844e-06,
-      "loss": 0.0864,
+      "loss": 0.0309,
       "step": 5680
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.459133500324707e-06,
-      "loss": 0.1371,
+      "loss": 0.0155,
       "step": 5681
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.457278040634567e-06,
-      "loss": 0.035,
+      "loss": 0.0911,
       "step": 5682
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.45542258094443e-06,
-      "loss": 0.034,
+      "loss": 0.0101,
       "step": 5683
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.453567121254291e-06,
-      "loss": 0.1923,
+      "loss": 0.159,
       "step": 5684
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.451711661564153e-06,
-      "loss": 0.1456,
+      "loss": 0.0633,
       "step": 5685
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.449856201874015e-06,
-      "loss": 0.2455,
+      "loss": 0.138,
       "step": 5686
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.448000742183877e-06,
-      "loss": 0.2432,
+      "loss": 0.0222,
       "step": 5687
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.446145282493739e-06,
-      "loss": 0.085,
+      "loss": 0.0174,
       "step": 5688
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.4442898228036e-06,
-      "loss": 0.1945,
+      "loss": 0.0625,
       "step": 5689
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.442434363113462e-06,
-      "loss": 0.19,
+      "loss": 0.0538,
       "step": 5690
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.440578903423324e-06,
-      "loss": 0.2514,
+      "loss": 0.0139,
       "step": 5691
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.438723443733186e-06,
-      "loss": 0.0833,
+      "loss": 0.0842,
       "step": 5692
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.436867984043048e-06,
-      "loss": 0.1452,
+      "loss": 0.0159,
       "step": 5693
     },
     {
       "epoch": 1.58,
       "learning_rate": 9.435012524352908e-06,
-      "loss": 0.0329,
+      "loss": 0.0152,
       "step": 5694
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.433157064662772e-06,
-      "loss": 0.1911,
+      "loss": 0.0853,
       "step": 5695
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.431301604972632e-06,
-      "loss": 0.1413,
+      "loss": 0.0119,
       "step": 5696
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.429446145282496e-06,
-      "loss": 0.0866,
+      "loss": 0.0175,
       "step": 5697
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.427590685592356e-06,
-      "loss": 0.136,
+      "loss": 0.0181,
       "step": 5698
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.425735225902218e-06,
-      "loss": 0.197,
+      "loss": 0.0933,
       "step": 5699
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.42387976621208e-06,
-      "loss": 0.142,
+      "loss": 0.0783,
       "step": 5700
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.422024306521941e-06,
-      "loss": 0.1393,
+      "loss": 0.1043,
       "step": 5701
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.420168846831803e-06,
-      "loss": 0.0319,
+      "loss": 0.1102,
       "step": 5702
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.418313387141665e-06,
-      "loss": 0.1927,
+      "loss": 0.012,
       "step": 5703
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.416457927451527e-06,
-      "loss": 0.2999,
+      "loss": 0.0232,
       "step": 5704
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.414602467761389e-06,
-      "loss": 0.0332,
+      "loss": 0.0633,
       "step": 5705
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.41274700807125e-06,
-      "loss": 0.1432,
+      "loss": 0.0823,
       "step": 5706
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.410891548381113e-06,
-      "loss": 0.2459,
+      "loss": 0.1161,
       "step": 5707
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.409036088690974e-06,
-      "loss": 0.1391,
+      "loss": 0.0678,
       "step": 5708
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.407180629000836e-06,
-      "loss": 0.1386,
+      "loss": 0.0537,
       "step": 5709
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.405325169310696e-06,
-      "loss": 0.1398,
+      "loss": 0.0937,
       "step": 5710
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.40346970962056e-06,
-      "loss": 0.1399,
+      "loss": 0.0121,
       "step": 5711
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.40161424993042e-06,
-      "loss": 0.1964,
+      "loss": 0.0503,
       "step": 5712
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.399758790240284e-06,
-      "loss": 0.1955,
+      "loss": 0.0222,
       "step": 5713
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.397903330550144e-06,
-      "loss": 0.1411,
+      "loss": 0.1577,
       "step": 5714
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.396047870860008e-06,
-      "loss": 0.1922,
+      "loss": 0.253,
       "step": 5715
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.394192411169868e-06,
-      "loss": 0.0336,
+      "loss": 0.1521,
       "step": 5716
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.39233695147973e-06,
-      "loss": 0.1921,
+      "loss": 0.2045,
       "step": 5717
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.390481491789591e-06,
-      "loss": 0.192,
+      "loss": 0.1018,
       "step": 5718
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.388626032099453e-06,
-      "loss": 0.1417,
+      "loss": 0.0151,
       "step": 5719
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.386770572409315e-06,
-      "loss": 0.1967,
+      "loss": 0.0489,
       "step": 5720
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.384915112719177e-06,
-      "loss": 0.0351,
+      "loss": 0.0198,
       "step": 5721
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.383059653029039e-06,
-      "loss": 0.0874,
+      "loss": 0.0614,
       "step": 5722
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.3812041933389e-06,
-      "loss": 0.0353,
+      "loss": 0.07,
       "step": 5723
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.379348733648763e-06,
-      "loss": 0.1406,
+      "loss": 0.0476,
       "step": 5724
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.377493273958625e-06,
-      "loss": 0.0877,
+      "loss": 0.0204,
       "step": 5725
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.375637814268486e-06,
-      "loss": 0.1921,
+      "loss": 0.0653,
       "step": 5726
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.373782354578348e-06,
-      "loss": 0.1903,
+      "loss": 0.1144,
       "step": 5727
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.371926894888208e-06,
-      "loss": 0.2986,
+      "loss": 0.1063,
       "step": 5728
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.370071435198072e-06,
-      "loss": 0.0861,
+      "loss": 0.1361,
       "step": 5729
     },
     {
       "epoch": 1.59,
       "learning_rate": 9.368215975507932e-06,
-      "loss": 0.1941,
+      "loss": 0.1616,
       "step": 5730
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.366360515817796e-06,
-      "loss": 0.1885,
+      "loss": 0.1342,
       "step": 5731
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.364505056127656e-06,
-      "loss": 0.1381,
+      "loss": 0.0664,
       "step": 5732
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.362649596437518e-06,
-      "loss": 0.0856,
+      "loss": 0.0566,
       "step": 5733
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.36079413674738e-06,
-      "loss": 0.141,
+      "loss": 0.0199,
       "step": 5734
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.358938677057242e-06,
-      "loss": 0.0346,
+      "loss": 0.0234,
       "step": 5735
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.357083217367103e-06,
-      "loss": 0.0882,
+      "loss": 0.026,
       "step": 5736
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.355227757676965e-06,
-      "loss": 0.3017,
+      "loss": 0.0776,
       "step": 5737
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.353372297986827e-06,
-      "loss": 0.0853,
+      "loss": 0.0477,
       "step": 5738
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.351516838296689e-06,
-      "loss": 0.1964,
+      "loss": 0.2427,
       "step": 5739
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.349661378606551e-06,
-      "loss": 0.0849,
+      "loss": 0.1326,
       "step": 5740
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.347805918916413e-06,
-      "loss": 0.2441,
+      "loss": 0.1326,
       "step": 5741
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.345950459226275e-06,
-      "loss": 0.1931,
+      "loss": 0.0702,
       "step": 5742
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.344094999536136e-06,
-      "loss": 0.0338,
+      "loss": 0.071,
       "step": 5743
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.342239539845998e-06,
-      "loss": 0.1941,
+      "loss": 0.0506,
       "step": 5744
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.34038408015586e-06,
-      "loss": 0.2449,
+      "loss": 0.1191,
       "step": 5745
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.33852862046572e-06,
-      "loss": 0.0887,
+      "loss": 0.1125,
       "step": 5746
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.336673160775584e-06,
-      "loss": 0.0336,
+      "loss": 0.122,
       "step": 5747
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.334817701085444e-06,
-      "loss": 0.086,
+      "loss": 0.2887,
       "step": 5748
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.332962241395308e-06,
-      "loss": 0.1928,
+      "loss": 0.0232,
       "step": 5749
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.331106781705168e-06,
-      "loss": 0.0869,
+      "loss": 0.1498,
       "step": 5750
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.32925132201503e-06,
-      "loss": 0.0873,
+      "loss": 0.176,
       "step": 5751
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.327395862324892e-06,
-      "loss": 0.0855,
+      "loss": 0.072,
       "step": 5752
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.325540402634754e-06,
-      "loss": 0.0867,
+      "loss": 0.1079,
       "step": 5753
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.323684942944615e-06,
-      "loss": 0.1414,
+      "loss": 0.0698,
       "step": 5754
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.321829483254477e-06,
-      "loss": 0.0854,
+      "loss": 0.0193,
       "step": 5755
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.319974023564339e-06,
-      "loss": 0.1401,
+      "loss": 0.0333,
       "step": 5756
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.318118563874201e-06,
-      "loss": 0.2487,
+      "loss": 0.0978,
       "step": 5757
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.316263104184063e-06,
-      "loss": 0.0856,
+      "loss": 0.0251,
       "step": 5758
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.314407644493925e-06,
-      "loss": 0.1383,
+      "loss": 0.0499,
       "step": 5759
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.312552184803787e-06,
-      "loss": 0.1947,
+      "loss": 0.0742,
       "step": 5760
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.310696725113648e-06,
-      "loss": 0.083,
+      "loss": 0.0525,
       "step": 5761
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.308841265423509e-06,
-      "loss": 0.1398,
+      "loss": 0.0417,
       "step": 5762
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.306985805733372e-06,
-      "loss": 0.086,
+      "loss": 0.2728,
       "step": 5763
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.305130346043232e-06,
-      "loss": 0.1429,
+      "loss": 0.2414,
       "step": 5764
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.303274886353096e-06,
-      "loss": 0.3541,
+      "loss": 0.055,
       "step": 5765
     },
     {
       "epoch": 1.6,
       "learning_rate": 9.301419426662956e-06,
-      "loss": 0.1402,
+      "loss": 0.0976,
       "step": 5766
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.299563966972818e-06,
-      "loss": 0.0872,
+      "loss": 0.0146,
       "step": 5767
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.29770850728268e-06,
-      "loss": 0.2434,
+      "loss": 0.1448,
       "step": 5768
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.295853047592542e-06,
-      "loss": 0.0844,
+      "loss": 0.021,
       "step": 5769
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.293997587902404e-06,
-      "loss": 0.1956,
+      "loss": 0.1452,
       "step": 5770
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.292142128212265e-06,
-      "loss": 0.1375,
+      "loss": 0.0436,
       "step": 5771
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.290286668522127e-06,
-      "loss": 0.1347,
+      "loss": 0.1126,
       "step": 5772
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.288431208831988e-06,
-      "loss": 0.1439,
+      "loss": 0.0177,
       "step": 5773
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.286575749141851e-06,
-      "loss": 0.0318,
+      "loss": 0.0797,
       "step": 5774
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.284720289451711e-06,
-      "loss": 0.1365,
+      "loss": 0.078,
       "step": 5775
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.282864829761575e-06,
-      "loss": 0.1425,
+      "loss": 0.0701,
       "step": 5776
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.281009370071435e-06,
-      "loss": 0.1881,
+      "loss": 0.0176,
       "step": 5777
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.279153910381299e-06,
-      "loss": 0.1917,
+      "loss": 0.0475,
       "step": 5778
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.277298450691159e-06,
-      "loss": 0.0879,
+      "loss": 0.0597,
       "step": 5779
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.27544299100102e-06,
-      "loss": 0.1365,
+      "loss": 0.0591,
       "step": 5780
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.273587531310882e-06,
-      "loss": 0.2448,
+      "loss": 0.0729,
       "step": 5781
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.271732071620744e-06,
-      "loss": 0.2409,
+      "loss": 0.0399,
       "step": 5782
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.269876611930606e-06,
-      "loss": 0.0852,
+      "loss": 0.0887,
       "step": 5783
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.268021152240468e-06,
-      "loss": 0.0871,
+      "loss": 0.1275,
       "step": 5784
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.26616569255033e-06,
-      "loss": 0.0838,
+      "loss": 0.073,
       "step": 5785
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.264310232860192e-06,
-      "loss": 0.0337,
+      "loss": 0.1068,
       "step": 5786
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.262454773170054e-06,
-      "loss": 0.2479,
+      "loss": 0.076,
       "step": 5787
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.260599313479916e-06,
-      "loss": 0.1384,
+      "loss": 0.1048,
       "step": 5788
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.258743853789777e-06,
-      "loss": 0.0834,
+      "loss": 0.0168,
       "step": 5789
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.25688839409964e-06,
-      "loss": 0.0853,
+      "loss": 0.0624,
       "step": 5790
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.2550329344095e-06,
-      "loss": 0.2431,
+      "loss": 0.1195,
       "step": 5791
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.253177474719363e-06,
-      "loss": 0.0872,
+      "loss": 0.0939,
       "step": 5792
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.251322015029223e-06,
-      "loss": 0.2427,
+      "loss": 0.0237,
       "step": 5793
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.249466555339087e-06,
-      "loss": 0.1954,
+      "loss": 0.231,
       "step": 5794
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.247611095648947e-06,
-      "loss": 0.141,
+      "loss": 0.0575,
       "step": 5795
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.245755635958809e-06,
-      "loss": 0.1902,
+      "loss": 0.1306,
       "step": 5796
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.24390017626867e-06,
-      "loss": 0.0872,
+      "loss": 0.0478,
       "step": 5797
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.242044716578533e-06,
-      "loss": 0.1399,
+      "loss": 0.1304,
       "step": 5798
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.240189256888394e-06,
-      "loss": 0.193,
+      "loss": 0.1358,
       "step": 5799
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.238333797198256e-06,
-      "loss": 0.1899,
+      "loss": 0.017,
       "step": 5800
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.236478337508118e-06,
-      "loss": 0.034,
+      "loss": 0.0489,
       "step": 5801
     },
     {
       "epoch": 1.61,
       "learning_rate": 9.23462287781798e-06,
-      "loss": 0.3006,
+      "loss": 0.1447,
       "step": 5802
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.232767418127842e-06,
-      "loss": 0.1906,
+      "loss": 0.0637,
       "step": 5803
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.230911958437704e-06,
-      "loss": 0.1932,
+      "loss": 0.0555,
       "step": 5804
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.229056498747566e-06,
-      "loss": 0.1395,
+      "loss": 0.0249,
       "step": 5805
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.227201039057428e-06,
-      "loss": 0.1895,
+      "loss": 0.1763,
       "step": 5806
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.22534557936729e-06,
-      "loss": 0.0346,
+      "loss": 0.2364,
       "step": 5807
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.223490119677151e-06,
-      "loss": 0.1881,
+      "loss": 0.0516,
       "step": 5808
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.221634659987011e-06,
-      "loss": 0.2968,
+      "loss": 0.0259,
       "step": 5809
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.219779200296875e-06,
-      "loss": 0.2439,
+      "loss": 0.1235,
       "step": 5810
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.217923740606735e-06,
-      "loss": 0.1901,
+      "loss": 0.1531,
       "step": 5811
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.216068280916599e-06,
-      "loss": 0.0377,
+      "loss": 0.0201,
       "step": 5812
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.214212821226459e-06,
-      "loss": 0.2346,
+      "loss": 0.0798,
       "step": 5813
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.21235736153632e-06,
-      "loss": 0.1898,
+      "loss": 0.0383,
       "step": 5814
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.210501901846183e-06,
-      "loss": 0.0384,
+      "loss": 0.1217,
       "step": 5815
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.208646442156045e-06,
-      "loss": 0.1417,
+      "loss": 0.1413,
       "step": 5816
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.206790982465906e-06,
-      "loss": 0.1854,
+      "loss": 0.0268,
       "step": 5817
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.204935522775768e-06,
-      "loss": 0.0889,
+      "loss": 0.0242,
       "step": 5818
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.20308006308563e-06,
-      "loss": 0.0396,
+      "loss": 0.092,
       "step": 5819
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.201224603395492e-06,
-      "loss": 0.191,
+      "loss": 0.0274,
       "step": 5820
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.199369143705354e-06,
-      "loss": 0.0869,
+      "loss": 0.0619,
       "step": 5821
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.197513684015216e-06,
-      "loss": 0.1405,
+      "loss": 0.0209,
       "step": 5822
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.195658224325078e-06,
-      "loss": 0.1945,
+      "loss": 0.0899,
       "step": 5823
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.19380276463494e-06,
-      "loss": 0.2966,
+      "loss": 0.1238,
       "step": 5824
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.1919473049448e-06,
-      "loss": 0.1925,
+      "loss": 0.0865,
       "step": 5825
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.190091845254663e-06,
-      "loss": 0.0892,
+      "loss": 0.0758,
       "step": 5826
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.188236385564523e-06,
-      "loss": 0.1917,
+      "loss": 0.159,
       "step": 5827
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.186380925874387e-06,
-      "loss": 0.1361,
+      "loss": 0.0242,
       "step": 5828
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.184525466184247e-06,
-      "loss": 0.0361,
+      "loss": 0.1257,
       "step": 5829
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.182670006494109e-06,
-      "loss": 0.0927,
+      "loss": 0.1147,
       "step": 5830
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.180814546803971e-06,
-      "loss": 0.2914,
+      "loss": 0.1261,
       "step": 5831
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.178959087113833e-06,
-      "loss": 0.2964,
+      "loss": 0.1173,
       "step": 5832
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.177103627423695e-06,
-      "loss": 0.2924,
+      "loss": 0.0722,
       "step": 5833
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.175248167733557e-06,
-      "loss": 0.09,
+      "loss": 0.056,
       "step": 5834
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.173392708043418e-06,
-      "loss": 0.19,
+      "loss": 0.0197,
       "step": 5835
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.17153724835328e-06,
-      "loss": 0.0872,
+      "loss": 0.1821,
       "step": 5836
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.169681788663142e-06,
-      "loss": 0.1373,
+      "loss": 0.054,
       "step": 5837
     },
     {
       "epoch": 1.62,
       "learning_rate": 9.167826328973004e-06,
-      "loss": 0.0381,
+      "loss": 0.0708,
       "step": 5838
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.165970869282866e-06,
-      "loss": 0.2409,
+      "loss": 0.0783,
       "step": 5839
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.164115409592728e-06,
-      "loss": 0.1919,
+      "loss": 0.0139,
       "step": 5840
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.16225994990259e-06,
-      "loss": 0.1892,
+      "loss": 0.0906,
       "step": 5841
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.160404490212452e-06,
-      "loss": 0.189,
+      "loss": 0.0186,
       "step": 5842
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.158549030522312e-06,
-      "loss": 0.2401,
+      "loss": 0.0186,
       "step": 5843
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.156693570832175e-06,
-      "loss": 0.1397,
+      "loss": 0.2157,
       "step": 5844
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.154838111142035e-06,
-      "loss": 0.0378,
+      "loss": 0.0529,
       "step": 5845
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.152982651451899e-06,
-      "loss": 0.2853,
+      "loss": 0.0612,
       "step": 5846
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.15112719176176e-06,
-      "loss": 0.1348,
+      "loss": 0.0237,
       "step": 5847
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.149271732071621e-06,
-      "loss": 0.3423,
+      "loss": 0.2207,
       "step": 5848
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.147416272381483e-06,
-      "loss": 0.0922,
+      "loss": 0.0391,
       "step": 5849
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.145560812691345e-06,
-      "loss": 0.09,
+      "loss": 0.0205,
       "step": 5850
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.143705353001207e-06,
-      "loss": 0.1896,
+      "loss": 0.0172,
       "step": 5851
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.141849893311069e-06,
-      "loss": 0.1915,
+      "loss": 0.0836,
       "step": 5852
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.13999443362093e-06,
-      "loss": 0.1422,
+      "loss": 0.0679,
       "step": 5853
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.138138973930792e-06,
-      "loss": 0.043,
+      "loss": 0.1575,
       "step": 5854
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.136283514240654e-06,
-      "loss": 0.1903,
+      "loss": 0.0205,
       "step": 5855
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.134428054550516e-06,
-      "loss": 0.0405,
+      "loss": 0.1362,
       "step": 5856
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.132572594860378e-06,
-      "loss": 0.1418,
+      "loss": 0.0572,
       "step": 5857
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.13071713517024e-06,
-      "loss": 0.1937,
+      "loss": 0.1341,
       "step": 5858
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.1288616754801e-06,
-      "loss": 0.1399,
+      "loss": 0.0155,
       "step": 5859
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.127006215789963e-06,
-      "loss": 0.1845,
+      "loss": 0.0272,
       "step": 5860
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.125150756099824e-06,
-      "loss": 0.0904,
+      "loss": 0.0692,
       "step": 5861
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.123295296409687e-06,
-      "loss": 0.1386,
+      "loss": 0.1559,
       "step": 5862
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.121439836719547e-06,
-      "loss": 0.1922,
+      "loss": 0.0269,
       "step": 5863
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.119584377029411e-06,
-      "loss": 0.1391,
+      "loss": 0.0262,
       "step": 5864
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.117728917339271e-06,
-      "loss": 0.1901,
+      "loss": 0.1289,
       "step": 5865
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.115873457649133e-06,
-      "loss": 0.2967,
+      "loss": 0.08,
       "step": 5866
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.114017997958995e-06,
-      "loss": 0.1881,
+      "loss": 0.0532,
       "step": 5867
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.112162538268857e-06,
-      "loss": 0.1929,
+      "loss": 0.0623,
       "step": 5868
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.110307078578719e-06,
-      "loss": 0.1402,
+      "loss": 0.2867,
       "step": 5869
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.10845161888858e-06,
-      "loss": 0.0874,
+      "loss": 0.0207,
       "step": 5870
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.106596159198442e-06,
-      "loss": 0.1925,
+      "loss": 0.0136,
       "step": 5871
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.104740699508304e-06,
-      "loss": 0.0893,
+      "loss": 0.1834,
       "step": 5872
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.102885239818166e-06,
-      "loss": 0.0893,
+      "loss": 0.1401,
       "step": 5873
     },
     {
       "epoch": 1.63,
       "learning_rate": 9.101029780128028e-06,
-      "loss": 0.086,
+      "loss": 0.0256,
       "step": 5874
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.09917432043789e-06,
-      "loss": 0.0366,
+      "loss": 0.0601,
       "step": 5875
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.097318860747752e-06,
-      "loss": 0.1907,
+      "loss": 0.0208,
       "step": 5876
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.095463401057612e-06,
-      "loss": 0.0364,
+      "loss": 0.1847,
       "step": 5877
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.093607941367475e-06,
-      "loss": 0.0872,
+      "loss": 0.0374,
       "step": 5878
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.091752481677336e-06,
-      "loss": 0.0875,
+      "loss": 0.0844,
       "step": 5879
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.0898970219872e-06,
-      "loss": 0.1412,
+      "loss": 0.0418,
       "step": 5880
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.08804156229706e-06,
-      "loss": 0.0348,
+      "loss": 0.0659,
       "step": 5881
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.086186102606921e-06,
-      "loss": 0.1932,
+      "loss": 0.1212,
       "step": 5882
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.084330642916783e-06,
-      "loss": 0.0862,
+      "loss": 0.0841,
       "step": 5883
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.082475183226645e-06,
-      "loss": 0.1948,
+      "loss": 0.1352,
       "step": 5884
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.080619723536507e-06,
-      "loss": 0.198,
+      "loss": 0.0248,
       "step": 5885
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.078764263846369e-06,
-      "loss": 0.188,
+      "loss": 0.1653,
       "step": 5886
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.07690880415623e-06,
-      "loss": 0.2475,
+      "loss": 0.1189,
       "step": 5887
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.075053344466092e-06,
-      "loss": 0.1369,
+      "loss": 0.0631,
       "step": 5888
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.073197884775954e-06,
-      "loss": 0.0307,
+      "loss": 0.098,
       "step": 5889
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.071342425085816e-06,
-      "loss": 0.1908,
+      "loss": 0.0939,
       "step": 5890
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.069486965395678e-06,
-      "loss": 0.0817,
+      "loss": 0.0661,
       "step": 5891
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.06763150570554e-06,
-      "loss": 0.1402,
+      "loss": 0.0354,
       "step": 5892
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.0657760460154e-06,
-      "loss": 0.138,
+      "loss": 0.1364,
       "step": 5893
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.063920586325264e-06,
-      "loss": 0.0295,
+      "loss": 0.0538,
       "step": 5894
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.062065126635124e-06,
-      "loss": 0.0861,
+      "loss": 0.0398,
       "step": 5895
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.060209666944987e-06,
-      "loss": 0.1353,
+      "loss": 0.0187,
       "step": 5896
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.058354207254848e-06,
-      "loss": 0.0857,
+      "loss": 0.0193,
       "step": 5897
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.056498747564711e-06,
-      "loss": 0.1972,
+      "loss": 0.0244,
       "step": 5898
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.054643287874571e-06,
-      "loss": 0.0864,
+      "loss": 0.0132,
       "step": 5899
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.052787828184433e-06,
-      "loss": 0.0861,
+      "loss": 0.0119,
       "step": 5900
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.050932368494295e-06,
-      "loss": 0.0853,
+      "loss": 0.0126,
       "step": 5901
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.049076908804157e-06,
-      "loss": 0.1968,
+      "loss": 0.1234,
       "step": 5902
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.047221449114019e-06,
-      "loss": 0.3031,
+      "loss": 0.1506,
       "step": 5903
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.04536598942388e-06,
-      "loss": 0.1359,
+      "loss": 0.0093,
       "step": 5904
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.043510529733743e-06,
-      "loss": 0.2529,
+      "loss": 0.1601,
       "step": 5905
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.041655070043604e-06,
-      "loss": 0.1382,
+      "loss": 0.11,
       "step": 5906
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.039799610353466e-06,
-      "loss": 0.08,
+      "loss": 0.0532,
       "step": 5907
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.037944150663328e-06,
-      "loss": 0.1387,
+      "loss": 0.0486,
       "step": 5908
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.03608869097319e-06,
-      "loss": 0.0293,
+      "loss": 0.0106,
       "step": 5909
     },
     {
       "epoch": 1.64,
       "learning_rate": 9.034233231283052e-06,
-      "loss": 0.1408,
+      "loss": 0.0729,
       "step": 5910
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.032377771592912e-06,
-      "loss": 0.0825,
+      "loss": 0.1139,
       "step": 5911
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.030522311902776e-06,
-      "loss": 0.085,
+      "loss": 0.0671,
       "step": 5912
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.028666852212636e-06,
-      "loss": 0.1353,
+      "loss": 0.0194,
       "step": 5913
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.0268113925225e-06,
-      "loss": 0.0842,
+      "loss": 0.1324,
       "step": 5914
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.02495593283236e-06,
-      "loss": 0.1388,
+      "loss": 0.17,
       "step": 5915
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.023100473142221e-06,
-      "loss": 0.0894,
+      "loss": 0.0844,
       "step": 5916
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.021245013452083e-06,
-      "loss": 0.3073,
+      "loss": 0.1143,
       "step": 5917
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.019389553761945e-06,
-      "loss": 0.029,
+      "loss": 0.0687,
       "step": 5918
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.017534094071807e-06,
-      "loss": 0.2501,
+      "loss": 0.0862,
       "step": 5919
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.015678634381669e-06,
-      "loss": 0.1376,
+      "loss": 0.0325,
       "step": 5920
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.01382317469153e-06,
-      "loss": 0.3673,
+      "loss": 0.0556,
       "step": 5921
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.011967715001391e-06,
-      "loss": 0.1943,
+      "loss": 0.1185,
       "step": 5922
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.010112255311255e-06,
-      "loss": 0.1371,
+      "loss": 0.0159,
       "step": 5923
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.008256795621115e-06,
-      "loss": 0.085,
+      "loss": 0.0456,
       "step": 5924
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.006401335930978e-06,
-      "loss": 0.1384,
+      "loss": 0.0228,
       "step": 5925
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.004545876240838e-06,
-      "loss": 0.1367,
+      "loss": 0.1028,
       "step": 5926
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.002690416550702e-06,
-      "loss": 0.1941,
+      "loss": 0.0243,
       "step": 5927
     },
     {
       "epoch": 1.65,
       "learning_rate": 9.000834956860562e-06,
-      "loss": 0.0843,
+      "loss": 0.0186,
       "step": 5928
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.998979497170424e-06,
-      "loss": 0.0885,
+      "loss": 0.0817,
       "step": 5929
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.997124037480286e-06,
-      "loss": 0.1375,
+      "loss": 0.0856,
       "step": 5930
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.995268577790148e-06,
-      "loss": 0.0826,
+      "loss": 0.2027,
       "step": 5931
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.99341311810001e-06,
-      "loss": 0.2465,
+      "loss": 0.1134,
       "step": 5932
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.991557658409872e-06,
-      "loss": 0.0869,
+      "loss": 0.0312,
       "step": 5933
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.989702198719733e-06,
-      "loss": 0.0856,
+      "loss": 0.0563,
       "step": 5934
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.987846739029595e-06,
-      "loss": 0.033,
+      "loss": 0.1322,
       "step": 5935
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.985991279339457e-06,
-      "loss": 0.0791,
+      "loss": 0.0907,
       "step": 5936
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.984135819649319e-06,
-      "loss": 0.0894,
+      "loss": 0.0556,
       "step": 5937
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.982280359959181e-06,
-      "loss": 0.0314,
+      "loss": 0.0853,
       "step": 5938
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.980424900269043e-06,
-      "loss": 0.1375,
+      "loss": 0.0817,
       "step": 5939
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.978569440578903e-06,
-      "loss": 0.0861,
+      "loss": 0.0139,
       "step": 5940
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.976713980888767e-06,
-      "loss": 0.0868,
+      "loss": 0.062,
       "step": 5941
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.974858521198627e-06,
-      "loss": 0.0854,
+      "loss": 0.0808,
       "step": 5942
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.97300306150849e-06,
-      "loss": 0.1395,
+      "loss": 0.0728,
       "step": 5943
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.97114760181835e-06,
-      "loss": 0.0288,
+      "loss": 0.0932,
       "step": 5944
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.969292142128212e-06,
-      "loss": 0.2972,
+      "loss": 0.0222,
       "step": 5945
     },
     {
       "epoch": 1.65,
       "learning_rate": 8.967436682438074e-06,
-      "loss": 0.1343,
+      "loss": 0.1377,
       "step": 5946
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.965581222747936e-06,
-      "loss": 0.1404,
+      "loss": 0.0521,
       "step": 5947
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.963725763057798e-06,
-      "loss": 0.252,
+      "loss": 0.1608,
       "step": 5948
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.96187030336766e-06,
-      "loss": 0.1858,
+      "loss": 0.0121,
       "step": 5949
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.960014843677522e-06,
-      "loss": 0.3078,
+      "loss": 0.1437,
       "step": 5950
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.958159383987384e-06,
-      "loss": 0.1974,
+      "loss": 0.0195,
       "step": 5951
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.956303924297245e-06,
-      "loss": 0.1422,
+      "loss": 0.1159,
       "step": 5952
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.954448464607107e-06,
-      "loss": 0.0831,
+      "loss": 0.0596,
       "step": 5953
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.95259300491697e-06,
-      "loss": 0.1416,
+      "loss": 0.0465,
       "step": 5954
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.950737545226831e-06,
-      "loss": 0.0855,
+      "loss": 0.0323,
       "step": 5955
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.948882085536691e-06,
-      "loss": 0.0861,
+      "loss": 0.0663,
       "step": 5956
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.947026625846555e-06,
-      "loss": 0.031,
+      "loss": 0.0992,
       "step": 5957
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.945171166156415e-06,
-      "loss": 0.0841,
+      "loss": 0.1126,
       "step": 5958
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.943315706466279e-06,
-      "loss": 0.0314,
+      "loss": 0.0106,
       "step": 5959
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.941460246776139e-06,
-      "loss": 0.1392,
+      "loss": 0.0203,
       "step": 5960
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.939604787086002e-06,
-      "loss": 0.1952,
+      "loss": 0.0124,
       "step": 5961
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.937749327395862e-06,
-      "loss": 0.1407,
+      "loss": 0.2419,
       "step": 5962
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.935893867705724e-06,
-      "loss": 0.1935,
+      "loss": 0.1848,
       "step": 5963
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.934038408015586e-06,
-      "loss": 0.1927,
+      "loss": 0.0431,
       "step": 5964
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.932182948325448e-06,
-      "loss": 0.0297,
+      "loss": 0.2379,
       "step": 5965
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.93032748863531e-06,
-      "loss": 0.2532,
+      "loss": 0.0939,
       "step": 5966
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.928472028945172e-06,
-      "loss": 0.1945,
+      "loss": 0.0161,
       "step": 5967
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.926616569255034e-06,
-      "loss": 0.0829,
+      "loss": 0.0681,
       "step": 5968
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.924761109564896e-06,
-      "loss": 0.3519,
+      "loss": 0.0643,
       "step": 5969
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.922905649874757e-06,
-      "loss": 0.0835,
+      "loss": 0.0659,
       "step": 5970
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.92105019018462e-06,
-      "loss": 0.087,
+      "loss": 0.0576,
       "step": 5971
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.919194730494481e-06,
-      "loss": 0.1899,
+      "loss": 0.1558,
       "step": 5972
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.917339270804343e-06,
-      "loss": 0.139,
+      "loss": 0.1042,
       "step": 5973
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.915483811114203e-06,
-      "loss": 0.193,
+      "loss": 0.0141,
       "step": 5974
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.913628351424067e-06,
-      "loss": 0.0854,
+      "loss": 0.1754,
       "step": 5975
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.911772891733927e-06,
-      "loss": 0.1967,
+      "loss": 0.1752,
       "step": 5976
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.90991743204379e-06,
-      "loss": 0.0838,
+      "loss": 0.0472,
       "step": 5977
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.90806197235365e-06,
-      "loss": 0.2453,
+      "loss": 0.0322,
       "step": 5978
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.906206512663513e-06,
-      "loss": 0.0857,
+      "loss": 0.0217,
       "step": 5979
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.904351052973374e-06,
-      "loss": 0.1376,
+      "loss": 0.0932,
       "step": 5980
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.902495593283236e-06,
-      "loss": 0.1318,
+      "loss": 0.0293,
       "step": 5981
     },
     {
       "epoch": 1.66,
       "learning_rate": 8.900640133593098e-06,
-      "loss": 0.1421,
+      "loss": 0.0348,
       "step": 5982
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.89878467390296e-06,
-      "loss": 0.194,
+      "loss": 0.0281,
       "step": 5983
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.896929214212822e-06,
-      "loss": 0.2451,
+      "loss": 0.2377,
       "step": 5984
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.895073754522684e-06,
-      "loss": 0.1401,
+      "loss": 0.0352,
       "step": 5985
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.893218294832546e-06,
-      "loss": 0.087,
+      "loss": 0.085,
       "step": 5986
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.891362835142407e-06,
-      "loss": 0.1948,
+      "loss": 0.0854,
       "step": 5987
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.88950737545227e-06,
-      "loss": 0.1985,
+      "loss": 0.0972,
       "step": 5988
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.887651915762131e-06,
-      "loss": 0.194,
+      "loss": 0.0147,
       "step": 5989
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.885796456071991e-06,
-      "loss": 0.1957,
+      "loss": 0.0573,
       "step": 5990
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.883940996381855e-06,
-      "loss": 0.0869,
+      "loss": 0.0711,
       "step": 5991
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.882085536691715e-06,
-      "loss": 0.1941,
+      "loss": 0.0272,
       "step": 5992
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.880230077001579e-06,
-      "loss": 0.1914,
+      "loss": 0.054,
       "step": 5993
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.878374617311439e-06,
-      "loss": 0.0865,
+      "loss": 0.082,
       "step": 5994
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.876519157621302e-06,
-      "loss": 0.1397,
+      "loss": 0.0542,
       "step": 5995
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.874663697931163e-06,
-      "loss": 0.1416,
+      "loss": 0.0503,
       "step": 5996
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.872808238241025e-06,
-      "loss": 0.1934,
+      "loss": 0.1146,
       "step": 5997
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.870952778550886e-06,
-      "loss": 0.2434,
+      "loss": 0.1378,
       "step": 5998
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.869097318860748e-06,
-      "loss": 0.2417,
+      "loss": 0.048,
       "step": 5999
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.86724185917061e-06,
-      "loss": 0.1379,
+      "loss": 0.0076,
       "step": 6000
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.865386399480472e-06,
-      "loss": 0.1908,
+      "loss": 0.0126,
       "step": 6001
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.863530939790334e-06,
-      "loss": 0.1377,
+      "loss": 0.0456,
       "step": 6002
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.861675480100196e-06,
-      "loss": 0.0369,
+      "loss": 0.048,
       "step": 6003
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.859820020410058e-06,
-      "loss": 0.0371,
+      "loss": 0.0737,
       "step": 6004
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.85796456071992e-06,
-      "loss": 0.0886,
+      "loss": 0.0158,
       "step": 6005
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.856109101029781e-06,
-      "loss": 0.0887,
+      "loss": 0.1056,
       "step": 6006
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.854253641339643e-06,
-      "loss": 0.1917,
+      "loss": 0.1436,
       "step": 6007
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.852398181649503e-06,
-      "loss": 0.0885,
+      "loss": 0.1481,
       "step": 6008
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.850542721959367e-06,
-      "loss": 0.2897,
+      "loss": 0.0582,
       "step": 6009
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.848687262269227e-06,
-      "loss": 0.0887,
+      "loss": 0.0599,
       "step": 6010
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.84683180257909e-06,
-      "loss": 0.1884,
+      "loss": 0.1248,
       "step": 6011
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.844976342888951e-06,
-      "loss": 0.191,
+      "loss": 0.0119,
       "step": 6012
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.843120883198813e-06,
-      "loss": 0.0872,
+      "loss": 0.0211,
       "step": 6013
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.841265423508675e-06,
-      "loss": 0.2871,
+      "loss": 0.0203,
       "step": 6014
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.839409963818536e-06,
-      "loss": 0.2434,
+      "loss": 0.1071,
       "step": 6015
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.837554504128398e-06,
-      "loss": 0.0372,
+      "loss": 0.1286,
       "step": 6016
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.83569904443826e-06,
-      "loss": 0.1405,
+      "loss": 0.111,
       "step": 6017
     },
     {
       "epoch": 1.67,
       "learning_rate": 8.833843584748122e-06,
-      "loss": 0.1421,
+      "loss": 0.111,
       "step": 6018
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.831988125057984e-06,
-      "loss": 0.1915,
+      "loss": 0.0778,
       "step": 6019
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.830132665367846e-06,
-      "loss": 0.2944,
+      "loss": 0.1456,
       "step": 6020
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.828277205677708e-06,
-      "loss": 0.1908,
+      "loss": 0.0216,
       "step": 6021
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.82642174598757e-06,
-      "loss": 0.1399,
+      "loss": 0.0556,
       "step": 6022
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.824566286297431e-06,
-      "loss": 0.0885,
+      "loss": 0.0354,
       "step": 6023
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.822710826607293e-06,
-      "loss": 0.2412,
+      "loss": 0.0468,
       "step": 6024
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.820855366917155e-06,
-      "loss": 0.2425,
+      "loss": 0.1534,
       "step": 6025
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.818999907227015e-06,
-      "loss": 0.0891,
+      "loss": 0.0257,
       "step": 6026
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.817144447536879e-06,
-      "loss": 0.188,
+      "loss": 0.0791,
       "step": 6027
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.815288987846739e-06,
-      "loss": 0.3397,
+      "loss": 0.0738,
       "step": 6028
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.813433528156603e-06,
-      "loss": 0.1378,
+      "loss": 0.057,
       "step": 6029
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.811578068466463e-06,
-      "loss": 0.1403,
+      "loss": 0.0922,
       "step": 6030
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.809722608776325e-06,
-      "loss": 0.086,
+      "loss": 0.1859,
       "step": 6031
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.807867149086187e-06,
-      "loss": 0.1444,
+      "loss": 0.1656,
       "step": 6032
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.806011689396048e-06,
-      "loss": 0.1393,
+      "loss": 0.1292,
       "step": 6033
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.80415622970591e-06,
-      "loss": 0.0885,
+      "loss": 0.1404,
       "step": 6034
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.802300770015772e-06,
-      "loss": 0.1363,
+      "loss": 0.1297,
       "step": 6035
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.800445310325634e-06,
-      "loss": 0.0888,
+      "loss": 0.0219,
       "step": 6036
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.798589850635496e-06,
-      "loss": 0.197,
+      "loss": 0.0171,
       "step": 6037
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.796734390945358e-06,
-      "loss": 0.3981,
+      "loss": 0.0149,
       "step": 6038
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.79487893125522e-06,
-      "loss": 0.0387,
+      "loss": 0.0666,
       "step": 6039
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.793023471565082e-06,
-      "loss": 0.1411,
+      "loss": 0.2291,
       "step": 6040
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.791168011874943e-06,
-      "loss": 0.1933,
+      "loss": 0.1507,
       "step": 6041
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.789312552184804e-06,
-      "loss": 0.0904,
+      "loss": 0.1048,
       "step": 6042
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.787457092494667e-06,
-      "loss": 0.0891,
+      "loss": 0.0554,
       "step": 6043
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.785601632804527e-06,
-      "loss": 0.0396,
+      "loss": 0.0185,
       "step": 6044
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.783746173114391e-06,
-      "loss": 0.0382,
+      "loss": 0.0501,
       "step": 6045
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.781890713424251e-06,
-      "loss": 0.1364,
+      "loss": 0.0173,
       "step": 6046
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.780035253734115e-06,
-      "loss": 0.1875,
+      "loss": 0.0172,
       "step": 6047
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.778179794043975e-06,
-      "loss": 0.1422,
+      "loss": 0.1516,
       "step": 6048
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.776324334353837e-06,
-      "loss": 0.0859,
+      "loss": 0.046,
       "step": 6049
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.774468874663699e-06,
-      "loss": 0.09,
+      "loss": 0.114,
       "step": 6050
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.77261341497356e-06,
-      "loss": 0.1416,
+      "loss": 0.0505,
       "step": 6051
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.770757955283422e-06,
-      "loss": 0.0861,
+      "loss": 0.1678,
       "step": 6052
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.768902495593284e-06,
-      "loss": 0.1874,
+      "loss": 0.0136,
       "step": 6053
     },
     {
       "epoch": 1.68,
       "learning_rate": 8.767047035903146e-06,
-      "loss": 0.0845,
+      "loss": 0.0219,
       "step": 6054
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.765191576213008e-06,
-      "loss": 0.189,
+      "loss": 0.054,
       "step": 6055
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.76333611652287e-06,
-      "loss": 0.1339,
+      "loss": 0.0137,
       "step": 6056
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.761480656832732e-06,
-      "loss": 0.1395,
+      "loss": 0.0601,
       "step": 6057
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.759625197142594e-06,
-      "loss": 0.0326,
+      "loss": 0.0174,
       "step": 6058
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.757769737452455e-06,
-      "loss": 0.0854,
+      "loss": 0.2369,
       "step": 6059
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.755914277762316e-06,
-      "loss": 0.0832,
+      "loss": 0.1626,
       "step": 6060
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.754058818072179e-06,
-      "loss": 0.2455,
+      "loss": 0.0476,
       "step": 6061
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.75220335838204e-06,
-      "loss": 0.086,
+      "loss": 0.2961,
       "step": 6062
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.750347898691903e-06,
-      "loss": 0.1375,
+      "loss": 0.0632,
       "step": 6063
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.748492439001763e-06,
-      "loss": 0.0829,
+      "loss": 0.0553,
       "step": 6064
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.746636979311625e-06,
-      "loss": 0.0839,
+      "loss": 0.0858,
       "step": 6065
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.744781519621487e-06,
-      "loss": 0.1935,
+      "loss": 0.0711,
       "step": 6066
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.742926059931349e-06,
-      "loss": 0.2449,
+      "loss": 0.0228,
       "step": 6067
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.74107060024121e-06,
-      "loss": 0.2455,
+      "loss": 0.0235,
       "step": 6068
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.739215140551072e-06,
-      "loss": 0.1924,
+      "loss": 0.0724,
       "step": 6069
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.737359680860934e-06,
-      "loss": 0.3018,
+      "loss": 0.0562,
       "step": 6070
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.735504221170796e-06,
-      "loss": 0.194,
+      "loss": 0.1602,
       "step": 6071
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.733648761480658e-06,
-      "loss": 0.0866,
+      "loss": 0.0807,
       "step": 6072
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.731793301790518e-06,
-      "loss": 0.4619,
+      "loss": 0.1736,
       "step": 6073
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.729937842100382e-06,
-      "loss": 0.1381,
+      "loss": 0.0266,
       "step": 6074
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.728082382410242e-06,
-      "loss": 0.1369,
+      "loss": 0.0528,
       "step": 6075
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.726226922720104e-06,
-      "loss": 0.0856,
+      "loss": 0.0878,
       "step": 6076
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.724371463029966e-06,
-      "loss": 0.0316,
+      "loss": 0.0785,
       "step": 6077
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.722516003339828e-06,
-      "loss": 0.1398,
+      "loss": 0.1841,
       "step": 6078
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.72066054364969e-06,
-      "loss": 0.0876,
+      "loss": 0.0271,
       "step": 6079
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.718805083959551e-06,
-      "loss": 0.2933,
+      "loss": 0.1162,
       "step": 6080
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.716949624269413e-06,
-      "loss": 0.1381,
+      "loss": 0.0983,
       "step": 6081
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.715094164579275e-06,
-      "loss": 0.0857,
+      "loss": 0.1106,
       "step": 6082
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.713238704889137e-06,
-      "loss": 0.1418,
+      "loss": 0.0885,
       "step": 6083
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.711383245198999e-06,
-      "loss": 0.245,
+      "loss": 0.1652,
       "step": 6084
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.70952778550886e-06,
-      "loss": 0.249,
+      "loss": 0.2024,
       "step": 6085
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.707672325818723e-06,
-      "loss": 0.0359,
+      "loss": 0.1076,
       "step": 6086
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.705816866128584e-06,
-      "loss": 0.0896,
+      "loss": 0.1154,
       "step": 6087
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.703961406438446e-06,
-      "loss": 0.1388,
+      "loss": 0.103,
       "step": 6088
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.702105946748306e-06,
-      "loss": 0.1409,
+      "loss": 0.0256,
       "step": 6089
     },
     {
       "epoch": 1.69,
       "learning_rate": 8.70025048705817e-06,
-      "loss": 0.2465,
+      "loss": 0.1117,
       "step": 6090
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.69839502736803e-06,
-      "loss": 0.0844,
+      "loss": 0.019,
       "step": 6091
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.696539567677894e-06,
-      "loss": 0.0855,
+      "loss": 0.0334,
       "step": 6092
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.694684107987754e-06,
-      "loss": 0.1906,
+      "loss": 0.0857,
       "step": 6093
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.692828648297616e-06,
-      "loss": 0.1933,
+      "loss": 0.1285,
       "step": 6094
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.690973188607478e-06,
-      "loss": 0.0346,
+      "loss": 0.0304,
       "step": 6095
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.68911772891734e-06,
-      "loss": 0.1392,
+      "loss": 0.0622,
       "step": 6096
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.687262269227201e-06,
-      "loss": 0.1399,
+      "loss": 0.0245,
       "step": 6097
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.685406809537063e-06,
-      "loss": 0.0876,
+      "loss": 0.0516,
       "step": 6098
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.683551349846925e-06,
-      "loss": 0.0887,
+      "loss": 0.063,
       "step": 6099
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.681695890156787e-06,
-      "loss": 0.0867,
+      "loss": 0.0919,
       "step": 6100
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.679840430466649e-06,
-      "loss": 0.1376,
+      "loss": 0.1158,
       "step": 6101
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.67798497077651e-06,
-      "loss": 0.0318,
+      "loss": 0.1089,
       "step": 6102
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.676129511086373e-06,
-      "loss": 0.0846,
+      "loss": 0.0738,
       "step": 6103
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.674274051396234e-06,
-      "loss": 0.032,
+      "loss": 0.0152,
       "step": 6104
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.672418591706095e-06,
-      "loss": 0.2447,
+      "loss": 0.1966,
       "step": 6105
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.670563132015958e-06,
-      "loss": 0.2419,
+      "loss": 0.0163,
       "step": 6106
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.668707672325818e-06,
-      "loss": 0.0325,
+      "loss": 0.1096,
       "step": 6107
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.666852212635682e-06,
-      "loss": 0.0849,
+      "loss": 0.0197,
       "step": 6108
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.664996752945542e-06,
-      "loss": 0.0854,
+      "loss": 0.0609,
       "step": 6109
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.663141293255404e-06,
-      "loss": 0.0845,
+      "loss": 0.0167,
       "step": 6110
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.661285833565266e-06,
-      "loss": 0.1943,
+      "loss": 0.0151,
       "step": 6111
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.659430373875128e-06,
-      "loss": 0.0827,
+      "loss": 0.0543,
       "step": 6112
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.65757491418499e-06,
-      "loss": 0.0288,
+      "loss": 0.1605,
       "step": 6113
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.655719454494852e-06,
-      "loss": 0.1344,
+      "loss": 0.0156,
       "step": 6114
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.653863994804713e-06,
-      "loss": 0.0849,
+      "loss": 0.0521,
       "step": 6115
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.652008535114575e-06,
-      "loss": 0.1936,
+      "loss": 0.0546,
       "step": 6116
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.650153075424437e-06,
-      "loss": 0.1959,
+      "loss": 0.1136,
       "step": 6117
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.648297615734299e-06,
-      "loss": 0.0284,
+      "loss": 0.0608,
       "step": 6118
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.646442156044161e-06,
-      "loss": 0.1949,
+      "loss": 0.061,
       "step": 6119
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.644586696354023e-06,
-      "loss": 0.1381,
+      "loss": 0.1414,
       "step": 6120
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.642731236663885e-06,
-      "loss": 0.0824,
+      "loss": 0.0885,
       "step": 6121
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.640875776973746e-06,
-      "loss": 0.1378,
+      "loss": 0.1395,
       "step": 6122
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.639020317283607e-06,
-      "loss": 0.1398,
+      "loss": 0.1016,
       "step": 6123
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.63716485759347e-06,
-      "loss": 0.0841,
+      "loss": 0.0718,
       "step": 6124
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.63530939790333e-06,
-      "loss": 0.0805,
+      "loss": 0.069,
       "step": 6125
     },
     {
       "epoch": 1.7,
       "learning_rate": 8.633453938213194e-06,
-      "loss": 0.0277,
+      "loss": 0.079,
       "step": 6126
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.631598478523054e-06,
-      "loss": 0.083,
+      "loss": 0.0803,
       "step": 6127
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.629743018832916e-06,
-      "loss": 0.3034,
+      "loss": 0.0645,
       "step": 6128
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.627887559142778e-06,
-      "loss": 0.0847,
+      "loss": 0.0496,
       "step": 6129
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.62603209945264e-06,
-      "loss": 0.2491,
+      "loss": 0.0197,
       "step": 6130
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.624176639762502e-06,
-      "loss": 0.0845,
+      "loss": 0.0749,
       "step": 6131
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.622321180072363e-06,
-      "loss": 0.1966,
+      "loss": 0.1213,
       "step": 6132
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.620465720382225e-06,
-      "loss": 0.1381,
+      "loss": 0.1232,
       "step": 6133
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.618610260692087e-06,
-      "loss": 0.142,
+      "loss": 0.1712,
       "step": 6134
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.616754801001949e-06,
-      "loss": 0.0269,
+      "loss": 0.2757,
       "step": 6135
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.614899341311811e-06,
-      "loss": 0.1919,
+      "loss": 0.0304,
       "step": 6136
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.613043881621673e-06,
-      "loss": 0.3649,
+      "loss": 0.1655,
       "step": 6137
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.611188421931535e-06,
-      "loss": 0.1942,
+      "loss": 0.08,
       "step": 6138
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.609332962241395e-06,
-      "loss": 0.0832,
+      "loss": 0.02,
       "step": 6139
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.607477502551258e-06,
-      "loss": 0.0834,
+      "loss": 0.0417,
       "step": 6140
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.605622042861119e-06,
-      "loss": 0.3023,
+      "loss": 0.0902,
       "step": 6141
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.603766583170982e-06,
-      "loss": 0.1933,
+      "loss": 0.1947,
       "step": 6142
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.601911123480842e-06,
-      "loss": 0.1422,
+      "loss": 0.0726,
       "step": 6143
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.600055663790706e-06,
-      "loss": 0.0295,
+      "loss": 0.151,
       "step": 6144
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.598200204100566e-06,
-      "loss": 0.2427,
+      "loss": 0.0841,
       "step": 6145
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.596344744410428e-06,
-      "loss": 0.1945,
+      "loss": 0.0314,
       "step": 6146
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.59448928472029e-06,
-      "loss": 0.0864,
+      "loss": 0.0248,
       "step": 6147
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.592633825030152e-06,
-      "loss": 0.1378,
+      "loss": 0.123,
       "step": 6148
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.590778365340014e-06,
-      "loss": 0.3562,
+      "loss": 0.0491,
       "step": 6149
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.588922905649875e-06,
-      "loss": 0.1404,
+      "loss": 0.0846,
       "step": 6150
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.587067445959737e-06,
-      "loss": 0.1905,
+      "loss": 0.0705,
       "step": 6151
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.5852119862696e-06,
-      "loss": 0.0822,
+      "loss": 0.0664,
       "step": 6152
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.583356526579461e-06,
-      "loss": 0.0337,
+      "loss": 0.0571,
       "step": 6153
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.581501066889323e-06,
-      "loss": 0.1929,
+      "loss": 0.0299,
       "step": 6154
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.579645607199185e-06,
-      "loss": 0.1363,
+      "loss": 0.136,
       "step": 6155
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.577790147509047e-06,
-      "loss": 0.1942,
+      "loss": 0.061,
       "step": 6156
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.575934687818907e-06,
-      "loss": 0.0338,
+      "loss": 0.0464,
       "step": 6157
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.57407922812877e-06,
-      "loss": 0.2937,
+      "loss": 0.0391,
       "step": 6158
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.57222376843863e-06,
-      "loss": 0.141,
+      "loss": 0.0221,
       "step": 6159
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.570368308748494e-06,
-      "loss": 0.2417,
+      "loss": 0.0302,
       "step": 6160
     },
     {
       "epoch": 1.71,
       "learning_rate": 8.568512849058354e-06,
-      "loss": 0.0341,
+      "loss": 0.0676,
       "step": 6161
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.566657389368216e-06,
-      "loss": 0.1414,
+      "loss": 0.0592,
       "step": 6162
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.564801929678078e-06,
-      "loss": 0.0902,
+      "loss": 0.0149,
       "step": 6163
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.56294646998794e-06,
-      "loss": 0.1433,
+      "loss": 0.0937,
       "step": 6164
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.561091010297802e-06,
-      "loss": 0.1381,
+      "loss": 0.024,
       "step": 6165
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.559235550607664e-06,
-      "loss": 0.2428,
+      "loss": 0.0715,
       "step": 6166
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.557380090917526e-06,
-      "loss": 0.088,
+      "loss": 0.1185,
       "step": 6167
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.555524631227387e-06,
-      "loss": 0.1392,
+      "loss": 0.016,
       "step": 6168
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.55366917153725e-06,
-      "loss": 0.0876,
+      "loss": 0.0179,
       "step": 6169
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.551813711847111e-06,
-      "loss": 0.0354,
+      "loss": 0.0117,
       "step": 6170
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.549958252156973e-06,
-      "loss": 0.1386,
+      "loss": 0.1293,
       "step": 6171
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.548102792466835e-06,
-      "loss": 0.2424,
+      "loss": 0.0524,
       "step": 6172
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.546247332776695e-06,
-      "loss": 0.193,
+      "loss": 0.169,
       "step": 6173
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.544391873086559e-06,
-      "loss": 0.0344,
+      "loss": 0.0183,
       "step": 6174
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.542536413396419e-06,
-      "loss": 0.0348,
+      "loss": 0.0585,
       "step": 6175
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.540680953706282e-06,
-      "loss": 0.0337,
+      "loss": 0.0638,
       "step": 6176
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.538825494016143e-06,
-      "loss": 0.2956,
+      "loss": 0.0501,
       "step": 6177
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.536970034326006e-06,
-      "loss": 0.0854,
+      "loss": 0.0093,
       "step": 6178
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.535114574635866e-06,
-      "loss": 0.1353,
+      "loss": 0.2466,
       "step": 6179
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.533259114945728e-06,
-      "loss": 0.0859,
+      "loss": 0.1259,
       "step": 6180
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.53140365525559e-06,
-      "loss": 0.1947,
+      "loss": 0.1255,
       "step": 6181
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.529548195565452e-06,
-      "loss": 0.1346,
+      "loss": 0.0697,
       "step": 6182
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.527692735875314e-06,
-      "loss": 0.1398,
+      "loss": 0.063,
       "step": 6183
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.525837276185176e-06,
-      "loss": 0.0807,
+      "loss": 0.0201,
       "step": 6184
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.523981816495038e-06,
-      "loss": 0.0334,
+      "loss": 0.0268,
       "step": 6185
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.5221263568049e-06,
-      "loss": 0.087,
+      "loss": 0.0521,
       "step": 6186
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.520270897114761e-06,
-      "loss": 0.2427,
+      "loss": 0.0678,
       "step": 6187
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.518415437424623e-06,
-      "loss": 0.031,
+      "loss": 0.0173,
       "step": 6188
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.516559977734485e-06,
-      "loss": 0.0312,
+      "loss": 0.117,
       "step": 6189
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.514704518044347e-06,
-      "loss": 0.1396,
+      "loss": 0.0211,
       "step": 6190
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.512849058354207e-06,
-      "loss": 0.2533,
+      "loss": 0.0667,
       "step": 6191
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.51099359866407e-06,
-      "loss": 0.3501,
+      "loss": 0.1244,
       "step": 6192
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.50913813897393e-06,
-      "loss": 0.086,
+      "loss": 0.0919,
       "step": 6193
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.507282679283794e-06,
-      "loss": 0.0305,
+      "loss": 0.0816,
       "step": 6194
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.505427219593655e-06,
-      "loss": 0.1937,
+      "loss": 0.0432,
       "step": 6195
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.503571759903516e-06,
-      "loss": 0.2481,
+      "loss": 0.1163,
       "step": 6196
     },
     {
       "epoch": 1.72,
       "learning_rate": 8.501716300213378e-06,
-      "loss": 0.1957,
+      "loss": 0.2656,
       "step": 6197
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.49986084052324e-06,
-      "loss": 0.1372,
+      "loss": 0.1276,
       "step": 6198
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.498005380833102e-06,
-      "loss": 0.3087,
+      "loss": 0.0522,
       "step": 6199
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.496149921142964e-06,
-      "loss": 0.0872,
+      "loss": 0.1066,
       "step": 6200
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.494294461452826e-06,
-      "loss": 0.031,
+      "loss": 0.0641,
       "step": 6201
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.492439001762688e-06,
-      "loss": 0.143,
+      "loss": 0.028,
       "step": 6202
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.49058354207255e-06,
-      "loss": 0.1399,
+      "loss": 0.0232,
       "step": 6203
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.488728082382411e-06,
-      "loss": 0.1455,
+      "loss": 0.1566,
       "step": 6204
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.486872622692273e-06,
-      "loss": 0.1926,
+      "loss": 0.0536,
       "step": 6205
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.485017163002135e-06,
-      "loss": 0.1968,
+      "loss": 0.097,
       "step": 6206
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.483161703311997e-06,
-      "loss": 0.1919,
+      "loss": 0.1033,
       "step": 6207
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.481306243621859e-06,
-      "loss": 0.139,
+      "loss": 0.0507,
       "step": 6208
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.479450783931719e-06,
-      "loss": 0.1919,
+      "loss": 0.0201,
       "step": 6209
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.477595324241583e-06,
-      "loss": 0.1388,
+      "loss": 0.0911,
       "step": 6210
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.475739864551443e-06,
-      "loss": 0.2477,
+      "loss": 0.0162,
       "step": 6211
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.473884404861306e-06,
-      "loss": 0.1425,
+      "loss": 0.054,
       "step": 6212
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.472028945171167e-06,
-      "loss": 0.0346,
+      "loss": 0.0714,
       "step": 6213
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.470173485481028e-06,
-      "loss": 0.1384,
+      "loss": 0.0871,
       "step": 6214
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.46831802579089e-06,
-      "loss": 0.0328,
+      "loss": 0.0453,
       "step": 6215
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.466462566100752e-06,
-      "loss": 0.0871,
+      "loss": 0.2225,
       "step": 6216
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.464607106410614e-06,
-      "loss": 0.2034,
+      "loss": 0.0951,
       "step": 6217
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.462751646720476e-06,
-      "loss": 0.0869,
+      "loss": 0.101,
       "step": 6218
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.460896187030338e-06,
-      "loss": 0.2974,
+      "loss": 0.0536,
       "step": 6219
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.4590407273402e-06,
-      "loss": 0.0854,
+      "loss": 0.1055,
       "step": 6220
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.457185267650061e-06,
-      "loss": 0.1432,
+      "loss": 0.0593,
       "step": 6221
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.455329807959923e-06,
-      "loss": 0.1955,
+      "loss": 0.0679,
       "step": 6222
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.453474348269785e-06,
-      "loss": 0.0332,
+      "loss": 0.1422,
       "step": 6223
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.451618888579647e-06,
-      "loss": 0.0324,
+      "loss": 0.1121,
       "step": 6224
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.449763428889507e-06,
-      "loss": 0.0332,
+      "loss": 0.0818,
       "step": 6225
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.44790796919937e-06,
-      "loss": 0.1391,
+      "loss": 0.0465,
       "step": 6226
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.446052509509231e-06,
-      "loss": 0.1907,
+      "loss": 0.1212,
       "step": 6227
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.444197049819093e-06,
-      "loss": 0.1911,
+      "loss": 0.1959,
       "step": 6228
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.442341590128955e-06,
-      "loss": 0.0336,
+      "loss": 0.1051,
       "step": 6229
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.440486130438817e-06,
-      "loss": 0.1421,
+      "loss": 0.0939,
       "step": 6230
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.438630670748679e-06,
-      "loss": 0.139,
+      "loss": 0.0253,
       "step": 6231
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.43677521105854e-06,
-      "loss": 0.0846,
+      "loss": 0.0989,
       "step": 6232
     },
     {
       "epoch": 1.73,
       "learning_rate": 8.434919751368402e-06,
-      "loss": 0.1416,
+      "loss": 0.0629,
       "step": 6233
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.433064291678264e-06,
-      "loss": 0.0307,
+      "loss": 0.0845,
       "step": 6234
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.431208831988126e-06,
-      "loss": 0.1367,
+      "loss": 0.1516,
       "step": 6235
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.429353372297986e-06,
-      "loss": 0.1447,
+      "loss": 0.0232,
       "step": 6236
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.42749791260785e-06,
-      "loss": 0.0825,
+      "loss": 0.058,
       "step": 6237
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.42564245291771e-06,
-      "loss": 0.0859,
+      "loss": 0.1766,
       "step": 6238
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.423786993227573e-06,
-      "loss": 0.2496,
+      "loss": 0.0235,
       "step": 6239
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.421931533537434e-06,
-      "loss": 0.0848,
+      "loss": 0.0233,
       "step": 6240
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.420076073847297e-06,
-      "loss": 0.1373,
+      "loss": 0.0505,
       "step": 6241
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.418220614157157e-06,
-      "loss": 0.029,
+      "loss": 0.0924,
       "step": 6242
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.41636515446702e-06,
-      "loss": 0.1954,
+      "loss": 0.1726,
       "step": 6243
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.414509694776881e-06,
-      "loss": 0.0293,
+      "loss": 0.0143,
       "step": 6244
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.412654235086743e-06,
-      "loss": 0.0859,
+      "loss": 0.1607,
       "step": 6245
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.410798775396605e-06,
-      "loss": 0.0814,
+      "loss": 0.1491,
       "step": 6246
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.408943315706467e-06,
-      "loss": 0.2479,
+      "loss": 0.1008,
       "step": 6247
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.407087856016329e-06,
-      "loss": 0.1957,
+      "loss": 0.1207,
       "step": 6248
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.40523239632619e-06,
-      "loss": 0.1374,
+      "loss": 0.0215,
       "step": 6249
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.403376936636052e-06,
-      "loss": 0.085,
+      "loss": 0.0744,
       "step": 6250
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.401521476945914e-06,
-      "loss": 0.0281,
+      "loss": 0.0613,
       "step": 6251
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.399666017255776e-06,
-      "loss": 0.0818,
+      "loss": 0.0197,
       "step": 6252
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.397810557565638e-06,
-      "loss": 0.0277,
+      "loss": 0.0463,
       "step": 6253
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.395955097875498e-06,
-      "loss": 0.1373,
+      "loss": 0.1207,
       "step": 6254
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.394099638185362e-06,
-      "loss": 0.0819,
+      "loss": 0.1115,
       "step": 6255
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.392244178495222e-06,
-      "loss": 0.2002,
+      "loss": 0.0185,
       "step": 6256
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.390388718805085e-06,
-      "loss": 0.0859,
+      "loss": 0.0222,
       "step": 6257
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.388533259114946e-06,
-      "loss": 0.1404,
+      "loss": 0.1049,
       "step": 6258
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.386677799424807e-06,
-      "loss": 0.0842,
+      "loss": 0.0354,
       "step": 6259
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.38482233973467e-06,
-      "loss": 0.0267,
+      "loss": 0.1417,
       "step": 6260
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.382966880044531e-06,
-      "loss": 0.1388,
+      "loss": 0.1269,
       "step": 6261
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.381111420354393e-06,
-      "loss": 0.2,
+      "loss": 0.0912,
       "step": 6262
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.379255960664255e-06,
-      "loss": 0.2508,
+      "loss": 0.1262,
       "step": 6263
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.377400500974117e-06,
-      "loss": 0.249,
+      "loss": 0.1716,
       "step": 6264
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.375545041283979e-06,
-      "loss": 0.0838,
+      "loss": 0.0161,
       "step": 6265
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.37368958159384e-06,
-      "loss": 0.1389,
+      "loss": 0.1953,
       "step": 6266
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.371834121903702e-06,
-      "loss": 0.3139,
+      "loss": 0.0425,
       "step": 6267
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.369978662213564e-06,
-      "loss": 0.1976,
+      "loss": 0.1064,
       "step": 6268
     },
     {
       "epoch": 1.74,
       "learning_rate": 8.368123202523426e-06,
-      "loss": 0.1989,
+      "loss": 0.0645,
       "step": 6269
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.366267742833288e-06,
-      "loss": 0.1963,
+      "loss": 0.0829,
       "step": 6270
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.36441228314315e-06,
-      "loss": 0.1369,
+      "loss": 0.0143,
       "step": 6271
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.36255682345301e-06,
-      "loss": 0.2457,
+      "loss": 0.061,
       "step": 6272
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.360701363762874e-06,
-      "loss": 0.1404,
+      "loss": 0.1415,
       "step": 6273
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.358845904072734e-06,
-      "loss": 0.0862,
+      "loss": 0.1718,
       "step": 6274
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.356990444382597e-06,
-      "loss": 0.141,
+      "loss": 0.021,
       "step": 6275
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.355134984692458e-06,
-      "loss": 0.141,
+      "loss": 0.0238,
       "step": 6276
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.35327952500232e-06,
-      "loss": 0.1992,
+      "loss": 0.0299,
       "step": 6277
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.351424065312181e-06,
-      "loss": 0.0296,
+      "loss": 0.0314,
       "step": 6278
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.349568605622043e-06,
-      "loss": 0.1884,
+      "loss": 0.0991,
       "step": 6279
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.347713145931905e-06,
-      "loss": 0.1428,
+      "loss": 0.0495,
       "step": 6280
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.345857686241767e-06,
-      "loss": 0.0877,
+      "loss": 0.11,
       "step": 6281
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.344002226551629e-06,
-      "loss": 0.0313,
+      "loss": 0.0539,
       "step": 6282
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.34214676686149e-06,
-      "loss": 0.1417,
+      "loss": 0.1591,
       "step": 6283
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.340291307171353e-06,
-      "loss": 0.142,
+      "loss": 0.0654,
       "step": 6284
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.338435847481214e-06,
-      "loss": 0.0845,
+      "loss": 0.0273,
       "step": 6285
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.336580387791076e-06,
-      "loss": 0.1408,
+      "loss": 0.0205,
       "step": 6286
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.334724928100938e-06,
-      "loss": 0.2479,
+      "loss": 0.0316,
       "step": 6287
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.332869468410798e-06,
-      "loss": 0.0865,
+      "loss": 0.1286,
       "step": 6288
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.331014008720662e-06,
-      "loss": 0.354,
+      "loss": 0.0786,
       "step": 6289
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.329158549030522e-06,
-      "loss": 0.2449,
+      "loss": 0.0108,
       "step": 6290
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.327303089340386e-06,
-      "loss": 0.2474,
+      "loss": 0.0638,
       "step": 6291
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.325447629650246e-06,
-      "loss": 0.4056,
+      "loss": 0.1192,
       "step": 6292
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.323592169960108e-06,
-      "loss": 0.0332,
+      "loss": 0.0127,
       "step": 6293
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.32173671026997e-06,
-      "loss": 0.1948,
+      "loss": 0.0958,
       "step": 6294
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.319881250579831e-06,
-      "loss": 0.1946,
+      "loss": 0.1022,
       "step": 6295
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.318025790889693e-06,
-      "loss": 0.1424,
+      "loss": 0.1281,
       "step": 6296
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.316170331199555e-06,
-      "loss": 0.0358,
+      "loss": 0.2301,
       "step": 6297
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.314314871509417e-06,
-      "loss": 0.1922,
+      "loss": 0.012,
       "step": 6298
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.312459411819279e-06,
-      "loss": 0.1895,
+      "loss": 0.0596,
       "step": 6299
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.31060395212914e-06,
-      "loss": 0.14,
+      "loss": 0.0139,
       "step": 6300
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.308748492439003e-06,
-      "loss": 0.1955,
+      "loss": 0.0925,
       "step": 6301
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.306893032748865e-06,
-      "loss": 0.0883,
+      "loss": 0.0131,
       "step": 6302
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.305037573058726e-06,
-      "loss": 0.2907,
+      "loss": 0.0326,
       "step": 6303
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.303182113368588e-06,
-      "loss": 0.0371,
+      "loss": 0.0434,
       "step": 6304
     },
     {
       "epoch": 1.75,
       "learning_rate": 8.30132665367845e-06,
-      "loss": 0.142,
+      "loss": 0.0535,
       "step": 6305
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.29947119398831e-06,
-      "loss": 0.0878,
+      "loss": 0.0828,
       "step": 6306
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.297615734298174e-06,
-      "loss": 0.297,
+      "loss": 0.1033,
       "step": 6307
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.295760274608034e-06,
-      "loss": 0.1897,
+      "loss": 0.0274,
       "step": 6308
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.293904814917898e-06,
-      "loss": 0.1361,
+      "loss": 0.0596,
       "step": 6309
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.292049355227758e-06,
-      "loss": 0.0383,
+      "loss": 0.0658,
       "step": 6310
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.29019389553762e-06,
-      "loss": 0.2393,
+      "loss": 0.1465,
       "step": 6311
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.288338435847482e-06,
-      "loss": 0.2382,
+      "loss": 0.072,
       "step": 6312
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.286482976157343e-06,
-      "loss": 0.038,
+      "loss": 0.0307,
       "step": 6313
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.284627516467205e-06,
-      "loss": 0.1392,
+      "loss": 0.1308,
       "step": 6314
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.282772056777067e-06,
-      "loss": 0.0379,
+      "loss": 0.1719,
       "step": 6315
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.280916597086929e-06,
-      "loss": 0.1387,
+      "loss": 0.1092,
       "step": 6316
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.279061137396791e-06,
-      "loss": 0.0372,
+      "loss": 0.0729,
       "step": 6317
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.277205677706653e-06,
-      "loss": 0.1882,
+      "loss": 0.056,
       "step": 6318
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.275350218016515e-06,
-      "loss": 0.1394,
+      "loss": 0.2216,
       "step": 6319
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.273494758326377e-06,
-      "loss": 0.1927,
+      "loss": 0.1152,
       "step": 6320
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.271639298636238e-06,
-      "loss": 0.0363,
+      "loss": 0.1222,
       "step": 6321
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.269783838946099e-06,
-      "loss": 0.0918,
+      "loss": 0.1881,
       "step": 6322
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.267928379255962e-06,
-      "loss": 0.0359,
+      "loss": 0.0702,
       "step": 6323
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.266072919565822e-06,
-      "loss": 0.1924,
+      "loss": 0.083,
       "step": 6324
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.264217459875686e-06,
-      "loss": 0.1388,
+      "loss": 0.0534,
       "step": 6325
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.262362000185546e-06,
-      "loss": 0.0343,
+      "loss": 0.1172,
       "step": 6326
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.26050654049541e-06,
-      "loss": 0.0856,
+      "loss": 0.0453,
       "step": 6327
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.25865108080527e-06,
-      "loss": 0.197,
+      "loss": 0.0839,
       "step": 6328
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.256795621115132e-06,
-      "loss": 0.1939,
+      "loss": 0.1782,
       "step": 6329
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.254940161424994e-06,
-      "loss": 0.1386,
+      "loss": 0.0494,
       "step": 6330
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.253084701734855e-06,
-      "loss": 0.1437,
+      "loss": 0.0692,
       "step": 6331
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.251229242044717e-06,
-      "loss": 0.0854,
+      "loss": 0.0588,
       "step": 6332
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.249373782354579e-06,
-      "loss": 0.138,
+      "loss": 0.0238,
       "step": 6333
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.247518322664441e-06,
-      "loss": 0.1389,
+      "loss": 0.0771,
       "step": 6334
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.245662862974303e-06,
-      "loss": 0.0311,
+      "loss": 0.1004,
       "step": 6335
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.243807403284165e-06,
-      "loss": 0.1955,
+      "loss": 0.0313,
       "step": 6336
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.241951943594027e-06,
-      "loss": 0.0855,
+      "loss": 0.1212,
       "step": 6337
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.240096483903888e-06,
-      "loss": 0.0317,
+      "loss": 0.1413,
       "step": 6338
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.23824102421375e-06,
-      "loss": 0.1947,
+      "loss": 0.0613,
       "step": 6339
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.23638556452361e-06,
-      "loss": 0.4115,
+      "loss": 0.0637,
       "step": 6340
     },
     {
       "epoch": 1.76,
       "learning_rate": 8.234530104833474e-06,
-      "loss": 0.3535,
+      "loss": 0.1145,
       "step": 6341
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.232674645143334e-06,
-      "loss": 0.1931,
+      "loss": 0.1964,
       "step": 6342
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.230819185453198e-06,
-      "loss": 0.0852,
+      "loss": 0.0781,
       "step": 6343
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.228963725763058e-06,
-      "loss": 0.0846,
+      "loss": 0.0563,
       "step": 6344
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.22710826607292e-06,
-      "loss": 0.0314,
+      "loss": 0.0869,
       "step": 6345
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.225252806382782e-06,
-      "loss": 0.0849,
+      "loss": 0.2002,
       "step": 6346
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.223397346692644e-06,
-      "loss": 0.0312,
+      "loss": 0.0339,
       "step": 6347
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.221541887002506e-06,
-      "loss": 0.1946,
+      "loss": 0.1768,
       "step": 6348
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.219686427312367e-06,
-      "loss": 0.1374,
+      "loss": 0.05,
       "step": 6349
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.21783096762223e-06,
-      "loss": 0.0876,
+      "loss": 0.0249,
       "step": 6350
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.215975507932091e-06,
-      "loss": 0.2478,
+      "loss": 0.0383,
       "step": 6351
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.214120048241953e-06,
-      "loss": 0.1912,
+      "loss": 0.0205,
       "step": 6352
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.212264588551815e-06,
-      "loss": 0.2436,
+      "loss": 0.1214,
       "step": 6353
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.210409128861677e-06,
-      "loss": 0.0309,
+      "loss": 0.128,
       "step": 6354
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.208553669171539e-06,
-      "loss": 0.1943,
+      "loss": 0.1043,
       "step": 6355
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.206698209481399e-06,
-      "loss": 0.1378,
+      "loss": 0.0881,
       "step": 6356
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.204842749791262e-06,
-      "loss": 0.306,
+      "loss": 0.0624,
       "step": 6357
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.202987290101123e-06,
-      "loss": 0.2492,
+      "loss": 0.1011,
       "step": 6358
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.201131830410986e-06,
-      "loss": 0.1385,
+      "loss": 0.0313,
       "step": 6359
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.199276370720846e-06,
-      "loss": 0.2448,
+      "loss": 0.0737,
       "step": 6360
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.19742091103071e-06,
-      "loss": 0.0859,
+      "loss": 0.0226,
       "step": 6361
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.19556545134057e-06,
-      "loss": 0.1423,
+      "loss": 0.0544,
       "step": 6362
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.193709991650432e-06,
-      "loss": 0.0342,
+      "loss": 0.0203,
       "step": 6363
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.191854531960294e-06,
-      "loss": 0.1934,
+      "loss": 0.0825,
       "step": 6364
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.189999072270156e-06,
-      "loss": 0.0854,
+      "loss": 0.0498,
       "step": 6365
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.188143612580017e-06,
-      "loss": 0.0806,
+      "loss": 0.0174,
       "step": 6366
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.18628815288988e-06,
-      "loss": 0.0836,
+      "loss": 0.0427,
       "step": 6367
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.184432693199741e-06,
-      "loss": 0.1402,
+      "loss": 0.0112,
       "step": 6368
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.182577233509603e-06,
-      "loss": 0.2487,
+      "loss": 0.1525,
       "step": 6369
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.180721773819465e-06,
-      "loss": 0.2989,
+      "loss": 0.0201,
       "step": 6370
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.178866314129327e-06,
-      "loss": 0.1957,
+      "loss": 0.1777,
       "step": 6371
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.177010854439189e-06,
-      "loss": 0.1375,
+      "loss": 0.1017,
       "step": 6372
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.17515539474905e-06,
-      "loss": 0.1888,
+      "loss": 0.0962,
       "step": 6373
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.17329993505891e-06,
-      "loss": 0.0847,
+      "loss": 0.0292,
       "step": 6374
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.171444475368774e-06,
-      "loss": 0.0863,
+      "loss": 0.0655,
       "step": 6375
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.169589015678634e-06,
-      "loss": 0.1933,
+      "loss": 0.0801,
       "step": 6376
     },
     {
       "epoch": 1.77,
       "learning_rate": 8.167733555988496e-06,
-      "loss": 0.0884,
+      "loss": 0.1785,
       "step": 6377
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.165878096298358e-06,
-      "loss": 0.2423,
+      "loss": 0.1499,
       "step": 6378
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.16402263660822e-06,
-      "loss": 0.1413,
+      "loss": 0.0334,
       "step": 6379
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.162167176918082e-06,
-      "loss": 0.1402,
+      "loss": 0.0103,
       "step": 6380
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.160311717227944e-06,
-      "loss": 0.2494,
+      "loss": 0.0615,
       "step": 6381
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.158456257537806e-06,
-      "loss": 0.089,
+      "loss": 0.019,
       "step": 6382
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.156600797847668e-06,
-      "loss": 0.1425,
+      "loss": 0.1839,
       "step": 6383
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.15474533815753e-06,
-      "loss": 0.1373,
+      "loss": 0.1024,
       "step": 6384
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.15288987846739e-06,
-      "loss": 0.0877,
+      "loss": 0.196,
       "step": 6385
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.151034418777253e-06,
-      "loss": 0.1399,
+      "loss": 0.0181,
       "step": 6386
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.149178959087113e-06,
-      "loss": 0.1433,
+      "loss": 0.1239,
       "step": 6387
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.147323499396977e-06,
-      "loss": 0.3461,
+      "loss": 0.0546,
       "step": 6388
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.145468039706837e-06,
-      "loss": 0.0852,
+      "loss": 0.0762,
       "step": 6389
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.1436125800167e-06,
-      "loss": 0.1383,
+      "loss": 0.0647,
       "step": 6390
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.141757120326561e-06,
-      "loss": 0.0919,
+      "loss": 0.0584,
       "step": 6391
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.139901660636423e-06,
-      "loss": 0.0367,
+      "loss": 0.1083,
       "step": 6392
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.138046200946285e-06,
-      "loss": 0.0894,
+      "loss": 0.0513,
       "step": 6393
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.136190741256146e-06,
-      "loss": 0.1433,
+      "loss": 0.0959,
       "step": 6394
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.134335281566008e-06,
-      "loss": 0.2413,
+      "loss": 0.0277,
       "step": 6395
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.13247982187587e-06,
-      "loss": 0.1372,
+      "loss": 0.0216,
       "step": 6396
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.130624362185732e-06,
-      "loss": 0.0867,
+      "loss": 0.0456,
       "step": 6397
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.128768902495594e-06,
-      "loss": 0.0351,
+      "loss": 0.115,
       "step": 6398
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.126913442805456e-06,
-      "loss": 0.1902,
+      "loss": 0.0996,
       "step": 6399
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.125057983115318e-06,
-      "loss": 0.1921,
+      "loss": 0.0676,
       "step": 6400
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.12320252342518e-06,
-      "loss": 0.244,
+      "loss": 0.0181,
       "step": 6401
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.121347063735041e-06,
-      "loss": 0.0854,
+      "loss": 0.0896,
       "step": 6402
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.119491604044902e-06,
-      "loss": 0.0871,
+      "loss": 0.0636,
       "step": 6403
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.117636144354765e-06,
-      "loss": 0.1368,
+      "loss": 0.1085,
       "step": 6404
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.115780684664625e-06,
-      "loss": 0.0848,
+      "loss": 0.0168,
       "step": 6405
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.113925224974489e-06,
-      "loss": 0.1424,
+      "loss": 0.0257,
       "step": 6406
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.112069765284349e-06,
-      "loss": 0.0851,
+      "loss": 0.1049,
       "step": 6407
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.110214305594211e-06,
-      "loss": 0.0843,
+      "loss": 0.0469,
       "step": 6408
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.108358845904073e-06,
-      "loss": 0.1938,
+      "loss": 0.0205,
       "step": 6409
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.106503386213935e-06,
-      "loss": 0.25,
+      "loss": 0.0835,
       "step": 6410
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.104647926523797e-06,
-      "loss": 0.0856,
+      "loss": 0.0094,
       "step": 6411
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.102792466833658e-06,
-      "loss": 0.0865,
+      "loss": 0.0192,
       "step": 6412
     },
     {
       "epoch": 1.78,
       "learning_rate": 8.10093700714352e-06,
-      "loss": 0.2482,
+      "loss": 0.1074,
       "step": 6413
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.099081547453382e-06,
-      "loss": 0.0866,
+      "loss": 0.0626,
       "step": 6414
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.097226087763244e-06,
-      "loss": 0.0875,
+      "loss": 0.1246,
       "step": 6415
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.095370628073106e-06,
-      "loss": 0.0868,
+      "loss": 0.1425,
       "step": 6416
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.093515168382968e-06,
-      "loss": 0.0862,
+      "loss": 0.0098,
       "step": 6417
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.09165970869283e-06,
-      "loss": 0.0854,
+      "loss": 0.2359,
       "step": 6418
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.08980424900269e-06,
-      "loss": 0.2445,
+      "loss": 0.0171,
       "step": 6419
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.087948789312553e-06,
-      "loss": 0.1954,
+      "loss": 0.0622,
       "step": 6420
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.086093329622414e-06,
-      "loss": 0.1356,
+      "loss": 0.094,
       "step": 6421
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.084237869932277e-06,
-      "loss": 0.139,
+      "loss": 0.0631,
       "step": 6422
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.082382410242137e-06,
-      "loss": 0.0829,
+      "loss": 0.0267,
       "step": 6423
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.080526950552001e-06,
-      "loss": 0.1352,
+      "loss": 0.0139,
       "step": 6424
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.078671490861861e-06,
-      "loss": 0.0843,
+      "loss": 0.0534,
       "step": 6425
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.076816031171723e-06,
-      "loss": 0.1963,
+      "loss": 0.0487,
       "step": 6426
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.074960571481585e-06,
-      "loss": 0.1873,
+      "loss": 0.0166,
       "step": 6427
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.073105111791447e-06,
-      "loss": 0.2981,
+      "loss": 0.04,
       "step": 6428
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.071249652101309e-06,
-      "loss": 0.141,
+      "loss": 0.1822,
       "step": 6429
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.06939419241117e-06,
-      "loss": 0.2999,
+      "loss": 0.0367,
       "step": 6430
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.067538732721032e-06,
-      "loss": 0.0858,
+      "loss": 0.0603,
       "step": 6431
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.065683273030894e-06,
-      "loss": 0.084,
+      "loss": 0.0561,
       "step": 6432
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.063827813340756e-06,
-      "loss": 0.031,
+      "loss": 0.0192,
       "step": 6433
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.061972353650618e-06,
-      "loss": 0.1386,
+      "loss": 0.1937,
       "step": 6434
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.06011689396048e-06,
-      "loss": 0.0327,
+      "loss": 0.1759,
       "step": 6435
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.058261434270342e-06,
-      "loss": 0.3016,
+      "loss": 0.0145,
       "step": 6436
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.056405974580202e-06,
-      "loss": 0.1367,
+      "loss": 0.0848,
       "step": 6437
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.054550514890065e-06,
-      "loss": 0.1358,
+      "loss": 0.0227,
       "step": 6438
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.052695055199926e-06,
-      "loss": 0.0866,
+      "loss": 0.0845,
       "step": 6439
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.050839595509789e-06,
-      "loss": 0.2457,
+      "loss": 0.0754,
       "step": 6440
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.04898413581965e-06,
-      "loss": 0.1892,
+      "loss": 0.0191,
       "step": 6441
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.047128676129511e-06,
-      "loss": 0.1393,
+      "loss": 0.019,
       "step": 6442
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.045273216439373e-06,
-      "loss": 0.0865,
+      "loss": 0.0179,
       "step": 6443
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.043417756749235e-06,
-      "loss": 0.1374,
+      "loss": 0.0804,
       "step": 6444
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.041562297059097e-06,
-      "loss": 0.4516,
+      "loss": 0.0572,
       "step": 6445
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.039706837368959e-06,
-      "loss": 0.189,
+      "loss": 0.1696,
       "step": 6446
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.03785137767882e-06,
-      "loss": 0.1438,
+      "loss": 0.0244,
       "step": 6447
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.035995917988682e-06,
-      "loss": 0.191,
+      "loss": 0.1089,
       "step": 6448
     },
     {
       "epoch": 1.79,
       "learning_rate": 8.034140458298544e-06,
-      "loss": 0.1405,
+      "loss": 0.0912,
       "step": 6449
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.032284998608406e-06,
-      "loss": 0.135,
+      "loss": 0.041,
       "step": 6450
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.030429538918268e-06,
-      "loss": 0.1927,
+      "loss": 0.0245,
       "step": 6451
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.02857407922813e-06,
-      "loss": 0.0852,
+      "loss": 0.0208,
       "step": 6452
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.026718619537992e-06,
-      "loss": 0.1946,
+      "loss": 0.0596,
       "step": 6453
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.024863159847854e-06,
-      "loss": 0.0883,
+      "loss": 0.0536,
       "step": 6454
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.023007700157714e-06,
-      "loss": 0.0863,
+      "loss": 0.0831,
       "step": 6455
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.021152240467577e-06,
-      "loss": 0.1892,
+      "loss": 0.0948,
       "step": 6456
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.019296780777438e-06,
-      "loss": 0.0889,
+      "loss": 0.0097,
       "step": 6457
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.017441321087301e-06,
-      "loss": 0.139,
+      "loss": 0.2536,
       "step": 6458
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.015585861397161e-06,
-      "loss": 0.0849,
+      "loss": 0.0683,
       "step": 6459
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.013730401707023e-06,
-      "loss": 0.1387,
+      "loss": 0.0758,
       "step": 6460
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.011874942016885e-06,
-      "loss": 0.352,
+      "loss": 0.0073,
       "step": 6461
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.010019482326747e-06,
-      "loss": 0.2374,
+      "loss": 0.1212,
       "step": 6462
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.008164022636609e-06,
-      "loss": 0.1884,
+      "loss": 0.1088,
       "step": 6463
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.00630856294647e-06,
-      "loss": 0.1822,
+      "loss": 0.1219,
       "step": 6464
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.004453103256333e-06,
-      "loss": 0.2403,
+      "loss": 0.0587,
       "step": 6465
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.002597643566194e-06,
-      "loss": 0.0864,
+      "loss": 0.0983,
       "step": 6466
     },
     {
       "epoch": 1.8,
       "learning_rate": 8.000742183876056e-06,
-      "loss": 0.0375,
+      "loss": 0.0082,
       "step": 6467
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.998886724185918e-06,
-      "loss": 0.0876,
+      "loss": 0.1787,
       "step": 6468
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.99703126449578e-06,
-      "loss": 0.1446,
+      "loss": 0.0973,
       "step": 6469
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.995175804805642e-06,
-      "loss": 0.14,
+      "loss": 0.0085,
       "step": 6470
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.993320345115502e-06,
-      "loss": 0.0905,
+      "loss": 0.1693,
       "step": 6471
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.991464885425366e-06,
-      "loss": 0.0865,
+      "loss": 0.1635,
       "step": 6472
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.989609425735226e-06,
-      "loss": 0.0362,
+      "loss": 0.1741,
       "step": 6473
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.98775396604509e-06,
-      "loss": 0.1449,
+      "loss": 0.0542,
       "step": 6474
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.98589850635495e-06,
-      "loss": 0.1409,
+      "loss": 0.1175,
       "step": 6475
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.984043046664811e-06,
-      "loss": 0.137,
+      "loss": 0.1195,
       "step": 6476
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.982187586974673e-06,
-      "loss": 0.1404,
+      "loss": 0.0241,
       "step": 6477
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.980332127284535e-06,
-      "loss": 0.0352,
+      "loss": 0.1084,
       "step": 6478
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.978476667594397e-06,
-      "loss": 0.0881,
+      "loss": 0.051,
       "step": 6479
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.976621207904259e-06,
-      "loss": 0.0895,
+      "loss": 0.0344,
       "step": 6480
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.97476574821412e-06,
-      "loss": 0.0849,
+      "loss": 0.065,
       "step": 6481
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.972910288523983e-06,
-      "loss": 0.0329,
+      "loss": 0.0411,
       "step": 6482
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.971054828833844e-06,
-      "loss": 0.192,
+      "loss": 0.0611,
       "step": 6483
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.969199369143706e-06,
-      "loss": 0.0859,
+      "loss": 0.0856,
       "step": 6484
     },
     {
       "epoch": 1.8,
       "learning_rate": 7.967343909453568e-06,
-      "loss": 0.1927,
+      "loss": 0.0285,
       "step": 6485
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.96548844976343e-06,
-      "loss": 0.0849,
+      "loss": 0.0763,
       "step": 6486
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.963632990073292e-06,
-      "loss": 0.1416,
+      "loss": 0.0601,
       "step": 6487
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.961777530383154e-06,
-      "loss": 0.1396,
+      "loss": 0.053,
       "step": 6488
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.959922070693014e-06,
-      "loss": 0.1361,
+      "loss": 0.0904,
       "step": 6489
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.958066611002878e-06,
-      "loss": 0.2974,
+      "loss": 0.0241,
       "step": 6490
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.956211151312738e-06,
-      "loss": 0.0824,
+      "loss": 0.0387,
       "step": 6491
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.954355691622601e-06,
-      "loss": 0.193,
+      "loss": 0.1194,
       "step": 6492
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.952500231932461e-06,
-      "loss": 0.1378,
+      "loss": 0.1266,
       "step": 6493
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.950644772242323e-06,
-      "loss": 0.0837,
+      "loss": 0.015,
       "step": 6494
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.948789312552185e-06,
-      "loss": 0.1907,
+      "loss": 0.072,
       "step": 6495
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.946933852862047e-06,
-      "loss": 0.0812,
+      "loss": 0.1128,
       "step": 6496
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.945078393171909e-06,
-      "loss": 0.1366,
+      "loss": 0.024,
       "step": 6497
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.94322293348177e-06,
-      "loss": 0.1989,
+      "loss": 0.0279,
       "step": 6498
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.941367473791633e-06,
-      "loss": 0.0859,
+      "loss": 0.2464,
       "step": 6499
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.939512014101495e-06,
-      "loss": 0.1327,
+      "loss": 0.0213,
       "step": 6500
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.937656554411356e-06,
-      "loss": 0.1426,
+      "loss": 0.06,
       "step": 6501
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.935801094721218e-06,
-      "loss": 0.1413,
+      "loss": 0.0175,
       "step": 6502
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.93394563503108e-06,
-      "loss": 0.1375,
+      "loss": 0.0935,
       "step": 6503
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.932090175340942e-06,
-      "loss": 0.2501,
+      "loss": 0.1575,
       "step": 6504
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.930234715650802e-06,
-      "loss": 0.305,
+      "loss": 0.0124,
       "step": 6505
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.928379255960666e-06,
-      "loss": 0.0822,
+      "loss": 0.0427,
       "step": 6506
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.926523796270526e-06,
-      "loss": 0.1387,
+      "loss": 0.1696,
       "step": 6507
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.92466833658039e-06,
-      "loss": 0.142,
+      "loss": 0.0158,
       "step": 6508
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.92281287689025e-06,
-      "loss": 0.2939,
+      "loss": 0.0175,
       "step": 6509
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.920957417200113e-06,
-      "loss": 0.0889,
+      "loss": 0.0142,
       "step": 6510
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.919101957509973e-06,
-      "loss": 0.031,
+      "loss": 0.2279,
       "step": 6511
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.917246497819835e-06,
-      "loss": 0.1955,
+      "loss": 0.1629,
       "step": 6512
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.915391038129697e-06,
-      "loss": 0.2984,
+      "loss": 0.0408,
       "step": 6513
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.913535578439559e-06,
-      "loss": 0.3524,
+      "loss": 0.0594,
       "step": 6514
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.911680118749421e-06,
-      "loss": 0.1398,
+      "loss": 0.0613,
       "step": 6515
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.909824659059283e-06,
-      "loss": 0.1959,
+      "loss": 0.0542,
       "step": 6516
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.907969199369145e-06,
-      "loss": 0.1378,
+      "loss": 0.0142,
       "step": 6517
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.906113739679007e-06,
-      "loss": 0.0856,
+      "loss": 0.2222,
       "step": 6518
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.904258279988868e-06,
-      "loss": 0.2482,
+      "loss": 0.154,
       "step": 6519
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.90240282029873e-06,
-      "loss": 0.1446,
+      "loss": 0.1044,
       "step": 6520
     },
     {
       "epoch": 1.81,
       "learning_rate": 7.900547360608592e-06,
-      "loss": 0.2355,
+      "loss": 0.146,
       "step": 6521
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.898691900918454e-06,
-      "loss": 0.0866,
+      "loss": 0.0613,
       "step": 6522
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.896836441228314e-06,
-      "loss": 0.1423,
+      "loss": 0.0625,
       "step": 6523
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.894980981538178e-06,
-      "loss": 0.0905,
+      "loss": 0.1265,
       "step": 6524
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.893125521848038e-06,
-      "loss": 0.0356,
+      "loss": 0.1213,
       "step": 6525
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.891270062157902e-06,
-      "loss": 0.0884,
+      "loss": 0.0613,
       "step": 6526
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.889414602467762e-06,
-      "loss": 0.1955,
+      "loss": 0.0503,
       "step": 6527
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.887559142777624e-06,
-      "loss": 0.1881,
+      "loss": 0.0476,
       "step": 6528
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.885703683087485e-06,
-      "loss": 0.1902,
+      "loss": 0.1075,
       "step": 6529
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.883848223397347e-06,
-      "loss": 0.4391,
+      "loss": 0.025,
       "step": 6530
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.88199276370721e-06,
-      "loss": 0.1931,
+      "loss": 0.1376,
       "step": 6531
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.880137304017071e-06,
-      "loss": 0.0383,
+      "loss": 0.0334,
       "step": 6532
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.878281844326933e-06,
-      "loss": 0.0893,
+      "loss": 0.0201,
       "step": 6533
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.876426384636793e-06,
-      "loss": 0.1405,
+      "loss": 0.1766,
       "step": 6534
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.874570924946657e-06,
-      "loss": 0.0375,
+      "loss": 0.0331,
       "step": 6535
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.872715465256517e-06,
-      "loss": 0.0911,
+      "loss": 0.0315,
       "step": 6536
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.87086000556638e-06,
-      "loss": 0.0367,
+      "loss": 0.0187,
       "step": 6537
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.86900454587624e-06,
-      "loss": 0.0373,
+      "loss": 0.1294,
       "step": 6538
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.867149086186102e-06,
-      "loss": 0.0877,
+      "loss": 0.1224,
       "step": 6539
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.865293626495964e-06,
-      "loss": 0.1938,
+      "loss": 0.0862,
       "step": 6540
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.863438166805826e-06,
-      "loss": 0.1869,
+      "loss": 0.1208,
       "step": 6541
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.861582707115688e-06,
-      "loss": 0.2472,
+      "loss": 0.0726,
       "step": 6542
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.85972724742555e-06,
-      "loss": 0.0884,
+      "loss": 0.0214,
       "step": 6543
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.857871787735412e-06,
-      "loss": 0.1407,
+      "loss": 0.023,
       "step": 6544
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.856016328045274e-06,
-      "loss": 0.1893,
+      "loss": 0.063,
       "step": 6545
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.854160868355136e-06,
-      "loss": 0.1912,
+      "loss": 0.0511,
       "step": 6546
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.852305408664997e-06,
-      "loss": 0.1394,
+      "loss": 0.0855,
       "step": 6547
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.85044994897486e-06,
-      "loss": 0.1917,
+      "loss": 0.0176,
       "step": 6548
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.848594489284721e-06,
-      "loss": 0.1369,
+      "loss": 0.1376,
       "step": 6549
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.846739029594583e-06,
-      "loss": 0.0841,
+      "loss": 0.0476,
       "step": 6550
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.844883569904445e-06,
-      "loss": 0.0881,
+      "loss": 0.1995,
       "step": 6551
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.843028110214305e-06,
-      "loss": 0.0337,
+      "loss": 0.0694,
       "step": 6552
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.841172650524169e-06,
-      "loss": 0.1357,
+      "loss": 0.0189,
       "step": 6553
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.839317190834029e-06,
-      "loss": 0.0885,
+      "loss": 0.1731,
       "step": 6554
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.837461731143892e-06,
-      "loss": 0.0837,
+      "loss": 0.1034,
       "step": 6555
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.835606271453753e-06,
-      "loss": 0.1956,
+      "loss": 0.1493,
       "step": 6556
     },
     {
       "epoch": 1.82,
       "learning_rate": 7.833750811763614e-06,
-      "loss": 0.1388,
+      "loss": 0.0314,
       "step": 6557
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.831895352073476e-06,
-      "loss": 0.0854,
+      "loss": 0.062,
       "step": 6558
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.830039892383338e-06,
-      "loss": 0.0885,
+      "loss": 0.0674,
       "step": 6559
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.8281844326932e-06,
-      "loss": 0.0852,
+      "loss": 0.0995,
       "step": 6560
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.826328973003062e-06,
-      "loss": 0.0868,
+      "loss": 0.0613,
       "step": 6561
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.824473513312924e-06,
-      "loss": 0.1925,
+      "loss": 0.0567,
       "step": 6562
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.822618053622786e-06,
-      "loss": 0.1354,
+      "loss": 0.0195,
       "step": 6563
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.820762593932648e-06,
-      "loss": 0.1942,
+      "loss": 0.0368,
       "step": 6564
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.81890713424251e-06,
-      "loss": 0.0838,
+      "loss": 0.0183,
       "step": 6565
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.817051674552371e-06,
-      "loss": 0.4723,
+      "loss": 0.0665,
       "step": 6566
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.815196214862233e-06,
-      "loss": 0.0838,
+      "loss": 0.1093,
       "step": 6567
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.813340755172093e-06,
-      "loss": 0.0305,
+      "loss": 0.0469,
       "step": 6568
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.811485295481957e-06,
-      "loss": 0.253,
+      "loss": 0.2039,
       "step": 6569
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.809629835791817e-06,
-      "loss": 0.0296,
+      "loss": 0.0205,
       "step": 6570
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.80777437610168e-06,
-      "loss": 0.1913,
+      "loss": 0.021,
       "step": 6571
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.80591891641154e-06,
-      "loss": 0.0848,
+      "loss": 0.0275,
       "step": 6572
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.804063456721404e-06,
-      "loss": 0.2463,
+      "loss": 0.0222,
       "step": 6573
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.802207997031265e-06,
-      "loss": 0.0831,
+      "loss": 0.0213,
       "step": 6574
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.800352537341126e-06,
-      "loss": 0.0844,
+      "loss": 0.1022,
       "step": 6575
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.798497077650988e-06,
-      "loss": 0.2498,
+      "loss": 0.0533,
       "step": 6576
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.79664161796085e-06,
-      "loss": 0.0855,
+      "loss": 0.0517,
       "step": 6577
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.794786158270712e-06,
-      "loss": 0.0834,
+      "loss": 0.1829,
       "step": 6578
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.792930698580574e-06,
-      "loss": 0.1377,
+      "loss": 0.0724,
       "step": 6579
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.791075238890436e-06,
-      "loss": 0.1375,
+      "loss": 0.019,
       "step": 6580
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.789219779200298e-06,
-      "loss": 0.0855,
+      "loss": 0.0102,
       "step": 6581
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.78736431951016e-06,
-      "loss": 0.0824,
+      "loss": 0.141,
       "step": 6582
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.785508859820021e-06,
-      "loss": 0.256,
+      "loss": 0.0777,
       "step": 6583
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.783653400129883e-06,
-      "loss": 0.1939,
+      "loss": 0.037,
       "step": 6584
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.781797940439745e-06,
-      "loss": 0.0819,
+      "loss": 0.1314,
       "step": 6585
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.779942480749605e-06,
-      "loss": 0.0866,
+      "loss": 0.075,
       "step": 6586
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.778087021059469e-06,
-      "loss": 0.0877,
+      "loss": 0.1301,
       "step": 6587
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.776231561369329e-06,
-      "loss": 0.1925,
+      "loss": 0.0512,
       "step": 6588
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.774376101679193e-06,
-      "loss": 0.0826,
+      "loss": 0.1129,
       "step": 6589
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.772520641989053e-06,
-      "loss": 0.0856,
+      "loss": 0.0429,
       "step": 6590
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.770665182298915e-06,
-      "loss": 0.1918,
+      "loss": 0.1537,
       "step": 6591
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.768809722608777e-06,
-      "loss": 0.194,
+      "loss": 0.0094,
       "step": 6592
     },
     {
       "epoch": 1.83,
       "learning_rate": 7.766954262918638e-06,
-      "loss": 0.188,
+      "loss": 0.1761,
       "step": 6593
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.7650988032285e-06,
-      "loss": 0.1375,
+      "loss": 0.0613,
       "step": 6594
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.763243343538362e-06,
-      "loss": 0.0303,
+      "loss": 0.0764,
       "step": 6595
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.761387883848224e-06,
-      "loss": 0.0853,
+      "loss": 0.0509,
       "step": 6596
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.759532424158086e-06,
-      "loss": 0.1977,
+      "loss": 0.1102,
       "step": 6597
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.757676964467948e-06,
-      "loss": 0.1398,
+      "loss": 0.0647,
       "step": 6598
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.75582150477781e-06,
-      "loss": 0.0302,
+      "loss": 0.0113,
       "step": 6599
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.753966045087671e-06,
-      "loss": 0.193,
+      "loss": 0.0812,
       "step": 6600
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.752110585397533e-06,
-      "loss": 0.3567,
+      "loss": 0.0223,
       "step": 6601
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.750255125707394e-06,
-      "loss": 0.0305,
+      "loss": 0.0512,
       "step": 6602
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.748399666017257e-06,
-      "loss": 0.1388,
+      "loss": 0.0147,
       "step": 6603
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.746544206327117e-06,
-      "loss": 0.1916,
+      "loss": 0.0159,
       "step": 6604
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.74468874663698e-06,
-      "loss": 0.1415,
+      "loss": 0.1801,
       "step": 6605
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.742833286946841e-06,
-      "loss": 0.1354,
+      "loss": 0.0785,
       "step": 6606
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.740977827256705e-06,
-      "loss": 0.0308,
+      "loss": 0.0769,
       "step": 6607
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.739122367566565e-06,
-      "loss": 0.1379,
+      "loss": 0.0556,
       "step": 6608
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.737266907876427e-06,
-      "loss": 0.1393,
+      "loss": 0.0205,
       "step": 6609
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.735411448186288e-06,
-      "loss": 0.1946,
+      "loss": 0.0247,
       "step": 6610
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.73355598849615e-06,
-      "loss": 0.1395,
+      "loss": 0.0456,
       "step": 6611
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.731700528806012e-06,
-      "loss": 0.1985,
+      "loss": 0.0948,
       "step": 6612
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.729845069115874e-06,
-      "loss": 0.0832,
+      "loss": 0.2907,
       "step": 6613
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.727989609425736e-06,
-      "loss": 0.1428,
+      "loss": 0.0702,
       "step": 6614
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.726134149735598e-06,
-      "loss": 0.0847,
+      "loss": 0.0467,
       "step": 6615
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.72427869004546e-06,
-      "loss": 0.1386,
+      "loss": 0.0132,
       "step": 6616
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.722423230355322e-06,
-      "loss": 0.2527,
+      "loss": 0.0572,
       "step": 6617
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.720567770665183e-06,
-      "loss": 0.0847,
+      "loss": 0.0162,
       "step": 6618
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.718712310975045e-06,
-      "loss": 0.1978,
+      "loss": 0.1163,
       "step": 6619
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.716856851284906e-06,
-      "loss": 0.2497,
+      "loss": 0.1446,
       "step": 6620
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.715001391594769e-06,
-      "loss": 0.0327,
+      "loss": 0.0891,
       "step": 6621
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.71314593190463e-06,
-      "loss": 0.2389,
+      "loss": 0.0971,
       "step": 6622
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.711290472214493e-06,
-      "loss": 0.0856,
+      "loss": 0.0845,
       "step": 6623
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.709435012524353e-06,
-      "loss": 0.1851,
+      "loss": 0.1042,
       "step": 6624
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.707579552834215e-06,
-      "loss": 0.1442,
+      "loss": 0.1445,
       "step": 6625
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.705724093144077e-06,
-      "loss": 0.0856,
+      "loss": 0.0783,
       "step": 6626
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.703868633453939e-06,
-      "loss": 0.0325,
+      "loss": 0.0534,
       "step": 6627
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.7020131737638e-06,
-      "loss": 0.0882,
+      "loss": 0.1025,
       "step": 6628
     },
     {
       "epoch": 1.84,
       "learning_rate": 7.700157714073662e-06,
-      "loss": 0.0864,
+      "loss": 0.1953,
       "step": 6629
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.698302254383524e-06,
-      "loss": 0.1401,
+      "loss": 0.0508,
       "step": 6630
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.696446794693386e-06,
-      "loss": 0.0857,
+      "loss": 0.0571,
       "step": 6631
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.694591335003248e-06,
-      "loss": 0.1911,
+      "loss": 0.0242,
       "step": 6632
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.69273587531311e-06,
-      "loss": 0.1393,
+      "loss": 0.0419,
       "step": 6633
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.690880415622972e-06,
-      "loss": 0.0831,
+      "loss": 0.0542,
       "step": 6634
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.689024955932834e-06,
-      "loss": 0.1953,
+      "loss": 0.0513,
       "step": 6635
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.687169496242695e-06,
-      "loss": 0.0317,
+      "loss": 0.038,
       "step": 6636
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.685314036552557e-06,
-      "loss": 0.0833,
+      "loss": 0.0302,
       "step": 6637
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.683458576862417e-06,
-      "loss": 0.1409,
+      "loss": 0.0596,
       "step": 6638
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.681603117172281e-06,
-      "loss": 0.1399,
+      "loss": 0.026,
       "step": 6639
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.679747657482141e-06,
-      "loss": 0.1372,
+      "loss": 0.0178,
       "step": 6640
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.677892197792005e-06,
-      "loss": 0.1426,
+      "loss": 0.0431,
       "step": 6641
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.676036738101865e-06,
-      "loss": 0.0835,
+      "loss": 0.156,
       "step": 6642
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.674181278411727e-06,
-      "loss": 0.14,
+      "loss": 0.1384,
       "step": 6643
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.672325818721589e-06,
-      "loss": 0.0302,
+      "loss": 0.0903,
       "step": 6644
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.67047035903145e-06,
-      "loss": 0.0843,
+      "loss": 0.0129,
       "step": 6645
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.668614899341312e-06,
-      "loss": 0.1396,
+      "loss": 0.0608,
       "step": 6646
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.666759439651174e-06,
-      "loss": 0.1381,
+      "loss": 0.1133,
       "step": 6647
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.664903979961036e-06,
-      "loss": 0.0851,
+      "loss": 0.0099,
       "step": 6648
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.663048520270898e-06,
-      "loss": 0.1362,
+      "loss": 0.0538,
       "step": 6649
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.66119306058076e-06,
-      "loss": 0.1398,
+      "loss": 0.086,
       "step": 6650
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.659337600890622e-06,
-      "loss": 0.1437,
+      "loss": 0.0184,
       "step": 6651
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.657482141200484e-06,
-      "loss": 0.1971,
+      "loss": 0.0572,
       "step": 6652
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.655626681510346e-06,
-      "loss": 0.0276,
+      "loss": 0.0403,
       "step": 6653
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.653771221820206e-06,
-      "loss": 0.1917,
+      "loss": 0.0933,
       "step": 6654
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.65191576213007e-06,
-      "loss": 0.1399,
+      "loss": 0.0134,
       "step": 6655
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.65006030243993e-06,
-      "loss": 0.1992,
+      "loss": 0.0465,
       "step": 6656
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.648204842749793e-06,
-      "loss": 0.3604,
+      "loss": 0.0724,
       "step": 6657
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.646349383059653e-06,
-      "loss": 0.0833,
+      "loss": 0.0261,
       "step": 6658
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.644493923369515e-06,
-      "loss": 0.2522,
+      "loss": 0.0201,
       "step": 6659
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.642638463679377e-06,
-      "loss": 0.136,
+      "loss": 0.018,
       "step": 6660
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.640783003989239e-06,
-      "loss": 0.086,
+      "loss": 0.0495,
       "step": 6661
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.6389275442991e-06,
-      "loss": 0.0844,
+      "loss": 0.0526,
       "step": 6662
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.637072084608963e-06,
-      "loss": 0.1416,
+      "loss": 0.0074,
       "step": 6663
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.635216624918824e-06,
-      "loss": 0.0299,
+      "loss": 0.0602,
       "step": 6664
     },
     {
       "epoch": 1.85,
       "learning_rate": 7.633361165228686e-06,
-      "loss": 0.1413,
+      "loss": 0.0767,
       "step": 6665
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.631505705538548e-06,
-      "loss": 0.0859,
+      "loss": 0.0471,
       "step": 6666
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.62965024584841e-06,
-      "loss": 0.1404,
+      "loss": 0.0132,
       "step": 6667
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.627794786158271e-06,
-      "loss": 0.0838,
+      "loss": 0.0432,
       "step": 6668
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.625939326468134e-06,
-      "loss": 0.2484,
+      "loss": 0.1041,
       "step": 6669
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.624083866777995e-06,
-      "loss": 0.2472,
+      "loss": 0.0261,
       "step": 6670
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.6222284070878575e-06,
-      "loss": 0.0294,
+      "loss": 0.0447,
       "step": 6671
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.6203729473977185e-06,
-      "loss": 0.1927,
+      "loss": 0.018,
       "step": 6672
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.61851748770758e-06,
-      "loss": 0.1367,
+      "loss": 0.0104,
       "step": 6673
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.616662028017442e-06,
-      "loss": 0.1398,
+      "loss": 0.0685,
       "step": 6674
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.614806568327304e-06,
-      "loss": 0.144,
+      "loss": 0.0865,
       "step": 6675
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.612951108637165e-06,
-      "loss": 0.1923,
+      "loss": 0.2324,
       "step": 6676
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.611095648947028e-06,
-      "loss": 0.1397,
+      "loss": 0.0701,
       "step": 6677
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.609240189256889e-06,
-      "loss": 0.0832,
+      "loss": 0.1287,
       "step": 6678
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.607384729566752e-06,
-      "loss": 0.1391,
+      "loss": 0.0684,
       "step": 6679
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.605529269876613e-06,
-      "loss": 0.0873,
+      "loss": 0.0694,
       "step": 6680
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.603673810186474e-06,
-      "loss": 0.0868,
+      "loss": 0.0782,
       "step": 6681
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.601818350496336e-06,
-      "loss": 0.1371,
+      "loss": 0.1078,
       "step": 6682
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.599962890806197e-06,
-      "loss": 0.1454,
+      "loss": 0.2107,
       "step": 6683
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.598107431116059e-06,
-      "loss": 0.0882,
+      "loss": 0.0509,
       "step": 6684
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.596251971425921e-06,
-      "loss": 0.0848,
+      "loss": 0.083,
       "step": 6685
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.594396511735783e-06,
-      "loss": 0.0308,
+      "loss": 0.0922,
       "step": 6686
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.592541052045644e-06,
-      "loss": 0.1347,
+      "loss": 0.0259,
       "step": 6687
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.590685592355507e-06,
-      "loss": 0.084,
+      "loss": 0.0508,
       "step": 6688
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.588830132665368e-06,
-      "loss": 0.0293,
+      "loss": 0.1157,
       "step": 6689
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.5869746729752305e-06,
-      "loss": 0.3055,
+      "loss": 0.091,
       "step": 6690
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.5851192132850915e-06,
-      "loss": 0.1429,
+      "loss": 0.0993,
       "step": 6691
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.583263753594953e-06,
-      "loss": 0.1421,
+      "loss": 0.1707,
       "step": 6692
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.581408293904815e-06,
-      "loss": 0.029,
+      "loss": 0.0533,
       "step": 6693
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.579552834214677e-06,
-      "loss": 0.0837,
+      "loss": 0.1614,
       "step": 6694
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.577697374524538e-06,
-      "loss": 0.1928,
+      "loss": 0.0229,
       "step": 6695
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.575841914834401e-06,
-      "loss": 0.2489,
+      "loss": 0.0576,
       "step": 6696
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.573986455144262e-06,
-      "loss": 0.084,
+      "loss": 0.1486,
       "step": 6697
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.572130995454125e-06,
-      "loss": 0.1413,
+      "loss": 0.0281,
       "step": 6698
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.570275535763986e-06,
-      "loss": 0.0885,
+      "loss": 0.0228,
       "step": 6699
     },
     {
       "epoch": 1.86,
       "learning_rate": 7.568420076073848e-06,
-      "loss": 0.0271,
+      "loss": 0.0259,
       "step": 6700
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.566564616383709e-06,
-      "loss": 0.1409,
+      "loss": 0.0404,
       "step": 6701
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.564709156693571e-06,
-      "loss": 0.1359,
+      "loss": 0.1218,
       "step": 6702
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.562853697003433e-06,
-      "loss": 0.2513,
+      "loss": 0.0291,
       "step": 6703
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.560998237313295e-06,
-      "loss": 0.1359,
+      "loss": 0.0329,
       "step": 6704
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.559142777623156e-06,
-      "loss": 0.1898,
+      "loss": 0.0592,
       "step": 6705
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.557287317933019e-06,
-      "loss": 0.1957,
+      "loss": 0.0258,
       "step": 6706
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.55543185824288e-06,
-      "loss": 0.0834,
+      "loss": 0.058,
       "step": 6707
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.5535763985527425e-06,
-      "loss": 0.0821,
+      "loss": 0.0694,
       "step": 6708
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.5517209388626035e-06,
-      "loss": 0.14,
+      "loss": 0.0947,
       "step": 6709
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.549865479172465e-06,
-      "loss": 0.0822,
+      "loss": 0.0903,
       "step": 6710
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.548010019482327e-06,
-      "loss": 0.1975,
+      "loss": 0.0415,
       "step": 6711
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.546154559792189e-06,
-      "loss": 0.3575,
+      "loss": 0.0156,
       "step": 6712
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.54429910010205e-06,
-      "loss": 0.2524,
+      "loss": 0.2835,
       "step": 6713
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.542443640411913e-06,
-      "loss": 0.0807,
+      "loss": 0.0423,
       "step": 6714
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.540588180721774e-06,
-      "loss": 0.0293,
+      "loss": 0.1079,
       "step": 6715
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.538732721031637e-06,
-      "loss": 0.0312,
+      "loss": 0.065,
       "step": 6716
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.536877261341498e-06,
-      "loss": 0.3057,
+      "loss": 0.0687,
       "step": 6717
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.5350218016513595e-06,
-      "loss": 0.1955,
+      "loss": 0.1151,
       "step": 6718
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.533166341961221e-06,
-      "loss": 0.088,
+      "loss": 0.0092,
       "step": 6719
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.531310882271083e-06,
-      "loss": 0.0843,
+      "loss": 0.0557,
       "step": 6720
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.529455422580944e-06,
-      "loss": 0.0837,
+      "loss": 0.0195,
       "step": 6721
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.527599962890807e-06,
-      "loss": 0.1418,
+      "loss": 0.0852,
       "step": 6722
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.525744503200668e-06,
-      "loss": 0.0869,
+      "loss": 0.0196,
       "step": 6723
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.523889043510531e-06,
-      "loss": 0.0853,
+      "loss": 0.0798,
       "step": 6724
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.522033583820392e-06,
-      "loss": 0.1884,
+      "loss": 0.0866,
       "step": 6725
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.5201781241302545e-06,
-      "loss": 0.2493,
+      "loss": 0.0916,
       "step": 6726
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.5183226644401155e-06,
-      "loss": 0.0311,
+      "loss": 0.0155,
       "step": 6727
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.516467204749977e-06,
-      "loss": 0.0845,
+      "loss": 0.078,
       "step": 6728
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.514611745059839e-06,
-      "loss": 0.247,
+      "loss": 0.0247,
       "step": 6729
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.512756285369701e-06,
-      "loss": 0.1422,
+      "loss": 0.0495,
       "step": 6730
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.510900825679562e-06,
-      "loss": 0.1366,
+      "loss": 0.1126,
       "step": 6731
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.509045365989425e-06,
-      "loss": 0.0849,
+      "loss": 0.0386,
       "step": 6732
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.507189906299286e-06,
-      "loss": 0.0318,
+      "loss": 0.0763,
       "step": 6733
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.505334446609149e-06,
-      "loss": 0.3515,
+      "loss": 0.1134,
       "step": 6734
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.50347898691901e-06,
-      "loss": 0.0857,
+      "loss": 0.0654,
       "step": 6735
     },
     {
       "epoch": 1.87,
       "learning_rate": 7.5016235272288715e-06,
-      "loss": 0.087,
+      "loss": 0.0691,
       "step": 6736
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.499768067538733e-06,
-      "loss": 0.0314,
+      "loss": 0.0112,
       "step": 6737
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.497912607848595e-06,
-      "loss": 0.1387,
+      "loss": 0.0096,
       "step": 6738
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.496057148158456e-06,
-      "loss": 0.1403,
+      "loss": 0.0201,
       "step": 6739
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.494201688468319e-06,
-      "loss": 0.1402,
+      "loss": 0.1281,
       "step": 6740
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.49234622877818e-06,
-      "loss": 0.0314,
+      "loss": 0.102,
       "step": 6741
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.490490769088043e-06,
-      "loss": 0.0859,
+      "loss": 0.1403,
       "step": 6742
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.488635309397904e-06,
-      "loss": 0.0832,
+      "loss": 0.15,
       "step": 6743
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.486779849707766e-06,
-      "loss": 0.1412,
+      "loss": 0.108,
       "step": 6744
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.4849243900176275e-06,
-      "loss": 0.0852,
+      "loss": 0.015,
       "step": 6745
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.483068930327489e-06,
-      "loss": 0.082,
+      "loss": 0.2128,
       "step": 6746
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.48121347063735e-06,
-      "loss": 0.1408,
+      "loss": 0.0197,
       "step": 6747
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.479358010947213e-06,
-      "loss": 0.1962,
+      "loss": 0.1466,
       "step": 6748
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.477502551257074e-06,
-      "loss": 0.0292,
+      "loss": 0.1267,
       "step": 6749
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.475647091566937e-06,
-      "loss": 0.2485,
+      "loss": 0.0525,
       "step": 6750
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.473791631876798e-06,
-      "loss": 0.0828,
+      "loss": 0.1051,
       "step": 6751
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.47193617218666e-06,
-      "loss": 0.1903,
+      "loss": 0.1159,
       "step": 6752
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.470080712496522e-06,
-      "loss": 0.1354,
+      "loss": 0.0547,
       "step": 6753
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.4682252528063835e-06,
-      "loss": 0.0279,
+      "loss": 0.1092,
       "step": 6754
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.4663697931162445e-06,
-      "loss": 0.087,
+      "loss": 0.1086,
       "step": 6755
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.464514333426107e-06,
-      "loss": 0.1972,
+      "loss": 0.0194,
       "step": 6756
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.462658873735968e-06,
-      "loss": 0.1375,
+      "loss": 0.0296,
       "step": 6757
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.460803414045831e-06,
-      "loss": 0.1994,
+      "loss": 0.1093,
       "step": 6758
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.458947954355692e-06,
-      "loss": 0.0843,
+      "loss": 0.181,
       "step": 6759
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.457092494665555e-06,
-      "loss": 0.1413,
+      "loss": 0.0242,
       "step": 6760
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.455237034975416e-06,
-      "loss": 0.0824,
+      "loss": 0.1154,
       "step": 6761
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.4533815752852776e-06,
-      "loss": 0.0276,
+      "loss": 0.0707,
       "step": 6762
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.4515261155951394e-06,
-      "loss": 0.2565,
+      "loss": 0.0927,
       "step": 6763
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.449670655905001e-06,
-      "loss": 0.1359,
+      "loss": 0.0541,
       "step": 6764
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.447815196214862e-06,
-      "loss": 0.19,
+      "loss": 0.058,
       "step": 6765
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.445959736524725e-06,
-      "loss": 0.1398,
+      "loss": 0.1224,
       "step": 6766
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.444104276834586e-06,
-      "loss": 0.087,
+      "loss": 0.032,
       "step": 6767
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.442248817144449e-06,
-      "loss": 0.1388,
+      "loss": 0.0538,
       "step": 6768
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.44039335745431e-06,
-      "loss": 0.1397,
+      "loss": 0.0901,
       "step": 6769
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.438537897764172e-06,
-      "loss": 0.3071,
+      "loss": 0.1056,
       "step": 6770
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.4366824380740336e-06,
-      "loss": 0.2494,
+      "loss": 0.0232,
       "step": 6771
     },
     {
       "epoch": 1.88,
       "learning_rate": 7.4348269783838954e-06,
-      "loss": 0.1345,
+      "loss": 0.0272,
       "step": 6772
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.4329715186937565e-06,
-      "loss": 0.0856,
+      "loss": 0.0202,
       "step": 6773
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.431116059003619e-06,
-      "loss": 0.1913,
+      "loss": 0.1556,
       "step": 6774
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.42926059931348e-06,
-      "loss": 0.1359,
+      "loss": 0.2659,
       "step": 6775
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.427405139623343e-06,
-      "loss": 0.1916,
+      "loss": 0.1091,
       "step": 6776
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.425549679933204e-06,
-      "loss": 0.0865,
+      "loss": 0.168,
       "step": 6777
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.423694220243066e-06,
-      "loss": 0.031,
+      "loss": 0.0668,
       "step": 6778
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.421838760552928e-06,
-      "loss": 0.25,
+      "loss": 0.0493,
       "step": 6779
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.4199833008627895e-06,
-      "loss": 0.03,
+      "loss": 0.1063,
       "step": 6780
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.4181278411726506e-06,
-      "loss": 0.1358,
+      "loss": 0.088,
       "step": 6781
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.416272381482513e-06,
-      "loss": 0.1898,
+      "loss": 0.1362,
       "step": 6782
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.414416921792374e-06,
-      "loss": 0.0869,
+      "loss": 0.0521,
       "step": 6783
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.412561462102237e-06,
-      "loss": 0.0855,
+      "loss": 0.0201,
       "step": 6784
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.410706002412098e-06,
-      "loss": 0.1425,
+      "loss": 0.0176,
       "step": 6785
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.408850542721961e-06,
-      "loss": 0.0843,
+      "loss": 0.1629,
       "step": 6786
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.406995083031822e-06,
-      "loss": 0.141,
+      "loss": 0.0699,
       "step": 6787
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.405139623341684e-06,
-      "loss": 0.0856,
+      "loss": 0.0935,
       "step": 6788
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.4032841636515455e-06,
-      "loss": 0.1373,
+      "loss": 0.1016,
       "step": 6789
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.401428703961407e-06,
-      "loss": 0.0846,
+      "loss": 0.0501,
       "step": 6790
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.3995732442712684e-06,
-      "loss": 0.0865,
+      "loss": 0.2092,
       "step": 6791
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.397717784581131e-06,
-      "loss": 0.0857,
+      "loss": 0.0757,
       "step": 6792
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.395862324890992e-06,
-      "loss": 0.2476,
+      "loss": 0.0581,
       "step": 6793
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.394006865200855e-06,
-      "loss": 0.1392,
+      "loss": 0.0867,
       "step": 6794
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.392151405510716e-06,
-      "loss": 0.0839,
+      "loss": 0.0287,
       "step": 6795
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.390295945820578e-06,
-      "loss": 0.1396,
+      "loss": 0.095,
       "step": 6796
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.38844048613044e-06,
-      "loss": 0.0847,
+      "loss": 0.0231,
       "step": 6797
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.3865850264403015e-06,
-      "loss": 0.1929,
+      "loss": 0.09,
       "step": 6798
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.3847295667501625e-06,
-      "loss": 0.2458,
+      "loss": 0.068,
       "step": 6799
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.382874107060025e-06,
-      "loss": 0.2466,
+      "loss": 0.0795,
       "step": 6800
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.381018647369886e-06,
-      "loss": 0.0297,
+      "loss": 0.0679,
       "step": 6801
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.379163187679749e-06,
-      "loss": 0.137,
+      "loss": 0.1196,
       "step": 6802
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.37730772798961e-06,
-      "loss": 0.1416,
+      "loss": 0.1831,
       "step": 6803
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.375452268299472e-06,
-      "loss": 0.2457,
+      "loss": 0.0247,
       "step": 6804
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.373596808609334e-06,
-      "loss": 0.1952,
+      "loss": 0.0532,
       "step": 6805
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.371741348919196e-06,
-      "loss": 0.0849,
+      "loss": 0.0171,
       "step": 6806
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.369885889229057e-06,
-      "loss": 0.1393,
+      "loss": 0.0856,
       "step": 6807
     },
     {
       "epoch": 1.89,
       "learning_rate": 7.368030429538919e-06,
-      "loss": 0.3011,
+      "loss": 0.031,
       "step": 6808
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.36617496984878e-06,
-      "loss": 0.0873,
+      "loss": 0.1244,
       "step": 6809
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.364319510158643e-06,
-      "loss": 0.1928,
+      "loss": 0.0695,
       "step": 6810
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.362464050468504e-06,
-      "loss": 0.1303,
+      "loss": 0.1594,
       "step": 6811
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.360608590778366e-06,
-      "loss": 0.0303,
+      "loss": 0.0642,
       "step": 6812
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.358753131088228e-06,
-      "loss": 0.0838,
+      "loss": 0.148,
       "step": 6813
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.35689767139809e-06,
-      "loss": 0.087,
+      "loss": 0.0485,
       "step": 6814
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.355042211707951e-06,
-      "loss": 0.086,
+      "loss": 0.0754,
       "step": 6815
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.3531867520178135e-06,
-      "loss": 0.0873,
+      "loss": 0.0812,
       "step": 6816
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.3513312923276745e-06,
-      "loss": 0.0325,
+      "loss": 0.0526,
       "step": 6817
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.349475832637537e-06,
-      "loss": 0.1421,
+      "loss": 0.0933,
       "step": 6818
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.347620372947398e-06,
-      "loss": 0.0319,
+      "loss": 0.2229,
       "step": 6819
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.345764913257261e-06,
-      "loss": 0.1385,
+      "loss": 0.1476,
       "step": 6820
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.343909453567122e-06,
-      "loss": 0.1385,
+      "loss": 0.109,
       "step": 6821
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.342053993876984e-06,
-      "loss": 0.0814,
+      "loss": 0.017,
       "step": 6822
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.340198534186846e-06,
-      "loss": 0.14,
+      "loss": 0.1035,
       "step": 6823
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.338343074496708e-06,
-      "loss": 0.0294,
+      "loss": 0.163,
       "step": 6824
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.336487614806569e-06,
-      "loss": 0.1935,
+      "loss": 0.0644,
       "step": 6825
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.334632155116431e-06,
-      "loss": 0.2,
+      "loss": 0.0979,
       "step": 6826
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.332776695426292e-06,
-      "loss": 0.0835,
+      "loss": 0.0758,
       "step": 6827
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.330921235736155e-06,
-      "loss": 0.0286,
+      "loss": 0.0246,
       "step": 6828
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.329065776046016e-06,
-      "loss": 0.1957,
+      "loss": 0.0607,
       "step": 6829
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.327210316355878e-06,
-      "loss": 0.0802,
+      "loss": 0.1158,
       "step": 6830
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.32535485666574e-06,
-      "loss": 0.1449,
+      "loss": 0.0131,
       "step": 6831
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.323499396975601e-06,
-      "loss": 0.141,
+      "loss": 0.0275,
       "step": 6832
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.321643937285463e-06,
-      "loss": 0.195,
+      "loss": 0.0151,
       "step": 6833
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.319788477595325e-06,
-      "loss": 0.307,
+      "loss": 0.1627,
       "step": 6834
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.3179330179051865e-06,
-      "loss": 0.1403,
+      "loss": 0.1298,
       "step": 6835
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.3160775582150475e-06,
-      "loss": 0.14,
+      "loss": 0.0868,
       "step": 6836
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.31422209852491e-06,
-      "loss": 0.2494,
+      "loss": 0.0211,
       "step": 6837
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.312366638834771e-06,
-      "loss": 0.0847,
+      "loss": 0.0547,
       "step": 6838
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.310511179144634e-06,
-      "loss": 0.2556,
+      "loss": 0.1733,
       "step": 6839
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.308655719454495e-06,
-      "loss": 0.1934,
+      "loss": 0.1128,
       "step": 6840
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.306800259764357e-06,
-      "loss": 0.0855,
+      "loss": 0.0506,
       "step": 6841
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.304944800074219e-06,
-      "loss": 0.0302,
+      "loss": 0.1132,
       "step": 6842
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.303089340384081e-06,
-      "loss": 0.0862,
+      "loss": 0.2408,
       "step": 6843
     },
     {
       "epoch": 1.9,
       "learning_rate": 7.301233880693942e-06,
-      "loss": 0.0879,
+      "loss": 0.0366,
       "step": 6844
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.299378421003804e-06,
-      "loss": 0.3556,
+      "loss": 0.0185,
       "step": 6845
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.297522961313665e-06,
-      "loss": 0.0304,
+      "loss": 0.0213,
       "step": 6846
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.295667501623528e-06,
-      "loss": 0.0308,
+      "loss": 0.0211,
       "step": 6847
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.293812041933389e-06,
-      "loss": 0.1372,
+      "loss": 0.173,
       "step": 6848
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.291956582243252e-06,
-      "loss": 0.1978,
+      "loss": 0.0928,
       "step": 6849
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.290101122553113e-06,
-      "loss": 0.1929,
+      "loss": 0.1208,
       "step": 6850
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.288245662862975e-06,
-      "loss": 0.1942,
+      "loss": 0.136,
       "step": 6851
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.286390203172837e-06,
-      "loss": 0.3061,
+      "loss": 0.0803,
       "step": 6852
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.2845347434826985e-06,
-      "loss": 0.1411,
+      "loss": 0.1316,
       "step": 6853
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.2826792837925595e-06,
-      "loss": 0.1888,
+      "loss": 0.1781,
       "step": 6854
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.280823824102422e-06,
-      "loss": 0.0869,
+      "loss": 0.0627,
       "step": 6855
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.278968364412283e-06,
-      "loss": 0.1942,
+      "loss": 0.0511,
       "step": 6856
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.277112904722146e-06,
-      "loss": 0.0308,
+      "loss": 0.0632,
       "step": 6857
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.275257445032007e-06,
-      "loss": 0.0319,
+      "loss": 0.0512,
       "step": 6858
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.273401985341869e-06,
-      "loss": 0.1411,
+      "loss": 0.0503,
       "step": 6859
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.271546525651731e-06,
-      "loss": 0.0855,
+      "loss": 0.0188,
       "step": 6860
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.269691065961593e-06,
-      "loss": 0.0881,
+      "loss": 0.0204,
       "step": 6861
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.267835606271454e-06,
-      "loss": 0.1425,
+      "loss": 0.0286,
       "step": 6862
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.265980146581316e-06,
-      "loss": 0.1406,
+      "loss": 0.0333,
       "step": 6863
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.264124686891177e-06,
-      "loss": 0.0847,
+      "loss": 0.0142,
       "step": 6864
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.26226922720104e-06,
-      "loss": 0.1887,
+      "loss": 0.0804,
       "step": 6865
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.260413767510901e-06,
-      "loss": 0.1431,
+      "loss": 0.0934,
       "step": 6866
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.258558307820763e-06,
-      "loss": 0.0841,
+      "loss": 0.1235,
       "step": 6867
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.256702848130625e-06,
-      "loss": 0.1356,
+      "loss": 0.0222,
       "step": 6868
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.254847388440487e-06,
-      "loss": 0.196,
+      "loss": 0.0161,
       "step": 6869
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.252991928750348e-06,
-      "loss": 0.0303,
+      "loss": 0.0144,
       "step": 6870
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.2511364690602105e-06,
-      "loss": 0.1921,
+      "loss": 0.0136,
       "step": 6871
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.2492810093700715e-06,
-      "loss": 0.0827,
+      "loss": 0.0122,
       "step": 6872
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.247425549679934e-06,
-      "loss": 0.0866,
+      "loss": 0.1828,
       "step": 6873
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.245570089989795e-06,
-      "loss": 0.1409,
+      "loss": 0.0166,
       "step": 6874
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.243714630299657e-06,
-      "loss": 0.084,
+      "loss": 0.1132,
       "step": 6875
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.241859170609519e-06,
-      "loss": 0.1391,
+      "loss": 0.0115,
       "step": 6876
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.240003710919381e-06,
-      "loss": 0.1936,
+      "loss": 0.0099,
       "step": 6877
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.238148251229242e-06,
-      "loss": 0.2012,
+      "loss": 0.0661,
       "step": 6878
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.2362927915391046e-06,
-      "loss": 0.0839,
+      "loss": 0.007,
       "step": 6879
     },
     {
       "epoch": 1.91,
       "learning_rate": 7.234437331848966e-06,
-      "loss": 0.2488,
+      "loss": 0.0488,
       "step": 6880
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.232581872158828e-06,
-      "loss": 0.1947,
+      "loss": 0.1517,
       "step": 6881
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.230726412468689e-06,
-      "loss": 0.2946,
+      "loss": 0.11,
       "step": 6882
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.228870952778552e-06,
-      "loss": 0.1919,
+      "loss": 0.0112,
       "step": 6883
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.227015493088413e-06,
-      "loss": 0.0878,
+      "loss": 0.0062,
       "step": 6884
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.225160033398275e-06,
-      "loss": 0.0302,
+      "loss": 0.0801,
       "step": 6885
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.223304573708137e-06,
-      "loss": 0.1421,
+      "loss": 0.0486,
       "step": 6886
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.221449114017999e-06,
-      "loss": 0.1363,
+      "loss": 0.0589,
       "step": 6887
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.21959365432786e-06,
-      "loss": 0.1883,
+      "loss": 0.0089,
       "step": 6888
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.2177381946377224e-06,
-      "loss": 0.1421,
+      "loss": 0.0106,
       "step": 6889
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.2158827349475835e-06,
-      "loss": 0.1952,
+      "loss": 0.1569,
       "step": 6890
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.214027275257446e-06,
-      "loss": 0.0843,
+      "loss": 0.0167,
       "step": 6891
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.212171815567307e-06,
-      "loss": 0.1939,
+      "loss": 0.0702,
       "step": 6892
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.210316355877169e-06,
-      "loss": 0.0854,
+      "loss": 0.0258,
       "step": 6893
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.208460896187031e-06,
-      "loss": 0.1391,
+      "loss": 0.1041,
       "step": 6894
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.206605436496893e-06,
-      "loss": 0.0877,
+      "loss": 0.0752,
       "step": 6895
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.204749976806754e-06,
-      "loss": 0.2461,
+      "loss": 0.0966,
       "step": 6896
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.2028945171166165e-06,
-      "loss": 0.085,
+      "loss": 0.0783,
       "step": 6897
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.2010390574264776e-06,
-      "loss": 0.0854,
+      "loss": 0.0094,
       "step": 6898
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.19918359773634e-06,
-      "loss": 0.1919,
+      "loss": 0.0598,
       "step": 6899
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.197328138046201e-06,
-      "loss": 0.0861,
+      "loss": 0.098,
       "step": 6900
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.195472678356063e-06,
-      "loss": 0.0868,
+      "loss": 0.2414,
       "step": 6901
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.193617218665925e-06,
-      "loss": 0.0868,
+      "loss": 0.0167,
       "step": 6902
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.191761758975787e-06,
-      "loss": 0.1426,
+      "loss": 0.1338,
       "step": 6903
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.189906299285648e-06,
-      "loss": 0.0324,
+      "loss": 0.1043,
       "step": 6904
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.188050839595511e-06,
-      "loss": 0.1394,
+      "loss": 0.1327,
       "step": 6905
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.186195379905372e-06,
-      "loss": 0.1393,
+      "loss": 0.0871,
       "step": 6906
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.184339920215234e-06,
-      "loss": 0.0855,
+      "loss": 0.0192,
       "step": 6907
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.1824844605250954e-06,
-      "loss": 0.0315,
+      "loss": 0.0929,
       "step": 6908
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.180629000834958e-06,
-      "loss": 0.031,
+      "loss": 0.0225,
       "step": 6909
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.178773541144819e-06,
-      "loss": 0.0852,
+      "loss": 0.0922,
       "step": 6910
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.176918081454681e-06,
-      "loss": 0.1391,
+      "loss": 0.081,
       "step": 6911
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.175062621764543e-06,
-      "loss": 0.0838,
+      "loss": 0.1059,
       "step": 6912
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.173207162074405e-06,
-      "loss": 0.1393,
+      "loss": 0.0295,
       "step": 6913
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.171351702384266e-06,
-      "loss": 0.0873,
+      "loss": 0.0501,
       "step": 6914
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.1694962426941285e-06,
-      "loss": 0.1422,
+      "loss": 0.1148,
       "step": 6915
     },
     {
       "epoch": 1.92,
       "learning_rate": 7.1676407830039895e-06,
-      "loss": 0.0865,
+      "loss": 0.0537,
       "step": 6916
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.165785323313852e-06,
-      "loss": 0.0289,
+      "loss": 0.1423,
       "step": 6917
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.163929863623713e-06,
-      "loss": 0.1383,
+      "loss": 0.0157,
       "step": 6918
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.162074403933575e-06,
-      "loss": 0.2503,
+      "loss": 0.0173,
       "step": 6919
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.160218944243437e-06,
-      "loss": 0.1977,
+      "loss": 0.0785,
       "step": 6920
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.158363484553299e-06,
-      "loss": 0.0822,
+      "loss": 0.0572,
       "step": 6921
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.15650802486316e-06,
-      "loss": 0.307,
+      "loss": 0.1974,
       "step": 6922
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.154652565173023e-06,
-      "loss": 0.1415,
+      "loss": 0.1486,
       "step": 6923
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.152797105482884e-06,
-      "loss": 0.0817,
+      "loss": 0.105,
       "step": 6924
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.150941645792746e-06,
-      "loss": 0.2512,
+      "loss": 0.0209,
       "step": 6925
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.149086186102607e-06,
-      "loss": 0.084,
+      "loss": 0.0553,
       "step": 6926
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.147230726412469e-06,
-      "loss": 0.1862,
+      "loss": 0.0602,
       "step": 6927
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.145375266722331e-06,
-      "loss": 0.0273,
+      "loss": 0.1567,
       "step": 6928
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.143519807032193e-06,
-      "loss": 0.1959,
+      "loss": 0.0183,
       "step": 6929
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.141664347342054e-06,
-      "loss": 0.3658,
+      "loss": 0.1417,
       "step": 6930
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.139808887651917e-06,
-      "loss": 0.1405,
+      "loss": 0.0225,
       "step": 6931
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.137953427961778e-06,
-      "loss": 0.1408,
+      "loss": 0.1459,
       "step": 6932
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.1360979682716405e-06,
-      "loss": 0.1349,
+      "loss": 0.148,
       "step": 6933
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.1342425085815015e-06,
-      "loss": 0.0861,
+      "loss": 0.0257,
       "step": 6934
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.132387048891363e-06,
-      "loss": 0.0828,
+      "loss": 0.0964,
       "step": 6935
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.130531589201225e-06,
-      "loss": 0.1923,
+      "loss": 0.0204,
       "step": 6936
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.128676129511087e-06,
-      "loss": 0.0294,
+      "loss": 0.0327,
       "step": 6937
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.126820669820948e-06,
-      "loss": 0.1914,
+      "loss": 0.1021,
       "step": 6938
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.124965210130811e-06,
-      "loss": 0.1371,
+      "loss": 0.0245,
       "step": 6939
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.123109750440672e-06,
-      "loss": 0.0838,
+      "loss": 0.0262,
       "step": 6940
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.121254290750535e-06,
-      "loss": 0.1416,
+      "loss": 0.0312,
       "step": 6941
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.119398831060396e-06,
-      "loss": 0.2536,
+      "loss": 0.054,
       "step": 6942
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.117543371370258e-06,
-      "loss": 0.2473,
+      "loss": 0.1043,
       "step": 6943
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.115687911680119e-06,
-      "loss": 0.0835,
+      "loss": 0.0652,
       "step": 6944
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.113832451989981e-06,
-      "loss": 0.0868,
+      "loss": 0.1324,
       "step": 6945
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.111976992299843e-06,
-      "loss": 0.1904,
+      "loss": 0.0473,
       "step": 6946
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.110121532609705e-06,
-      "loss": 0.1941,
+      "loss": 0.2169,
       "step": 6947
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.108266072919566e-06,
-      "loss": 0.1425,
+      "loss": 0.0571,
       "step": 6948
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.106410613229429e-06,
-      "loss": 0.1359,
+      "loss": 0.0825,
       "step": 6949
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.10455515353929e-06,
-      "loss": 0.1919,
+      "loss": 0.0566,
       "step": 6950
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.1026996938491525e-06,
-      "loss": 0.1918,
+      "loss": 0.0571,
       "step": 6951
     },
     {
       "epoch": 1.93,
       "learning_rate": 7.1008442341590135e-06,
-      "loss": 0.1405,
+      "loss": 0.0709,
       "step": 6952
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.098988774468875e-06,
-      "loss": 0.245,
+      "loss": 0.1486,
       "step": 6953
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.097133314778737e-06,
-      "loss": 0.0857,
+      "loss": 0.049,
       "step": 6954
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.095277855088599e-06,
-      "loss": 0.1359,
+      "loss": 0.1596,
       "step": 6955
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.09342239539846e-06,
-      "loss": 0.0846,
+      "loss": 0.0976,
       "step": 6956
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.091566935708323e-06,
-      "loss": 0.0331,
+      "loss": 0.0181,
       "step": 6957
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.089711476018184e-06,
-      "loss": 0.1417,
+      "loss": 0.0799,
       "step": 6958
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.087856016328047e-06,
-      "loss": 0.0328,
+      "loss": 0.0223,
       "step": 6959
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.086000556637908e-06,
-      "loss": 0.032,
+      "loss": 0.0926,
       "step": 6960
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.0841450969477695e-06,
-      "loss": 0.0813,
+      "loss": 0.0669,
       "step": 6961
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.082289637257631e-06,
-      "loss": 0.0861,
+      "loss": 0.0275,
       "step": 6962
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.080434177567493e-06,
-      "loss": 0.243,
+      "loss": 0.0966,
       "step": 6963
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.078578717877354e-06,
-      "loss": 0.0319,
+      "loss": 0.1274,
       "step": 6964
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.076723258187217e-06,
-      "loss": 0.1911,
+      "loss": 0.0128,
       "step": 6965
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.074867798497078e-06,
-      "loss": 0.3037,
+      "loss": 0.1063,
       "step": 6966
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.073012338806941e-06,
-      "loss": 0.088,
+      "loss": 0.0097,
       "step": 6967
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.071156879116802e-06,
-      "loss": 0.3069,
+      "loss": 0.1924,
       "step": 6968
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.0693014194266645e-06,
-      "loss": 0.3036,
+      "loss": 0.1722,
       "step": 6969
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.0674459597365255e-06,
-      "loss": 0.1887,
+      "loss": 0.0563,
       "step": 6970
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.065590500046387e-06,
-      "loss": 0.343,
+      "loss": 0.0573,
       "step": 6971
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.063735040356249e-06,
-      "loss": 0.1927,
+      "loss": 0.213,
       "step": 6972
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.061879580666111e-06,
-      "loss": 0.2438,
+      "loss": 0.0714,
       "step": 6973
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.060024120975972e-06,
-      "loss": 0.0332,
+      "loss": 0.0722,
       "step": 6974
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.058168661285835e-06,
-      "loss": 0.035,
+      "loss": 0.0488,
       "step": 6975
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.056313201595696e-06,
-      "loss": 0.0857,
+      "loss": 0.0182,
       "step": 6976
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.0544577419055586e-06,
-      "loss": 0.0882,
+      "loss": 0.0916,
       "step": 6977
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.05260228221542e-06,
-      "loss": 0.1396,
+      "loss": 0.1353,
       "step": 6978
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.0507468225252815e-06,
-      "loss": 0.0885,
+      "loss": 0.0234,
       "step": 6979
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.048891362835143e-06,
-      "loss": 0.296,
+      "loss": 0.0563,
       "step": 6980
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.047035903145005e-06,
-      "loss": 0.2447,
+      "loss": 0.0234,
       "step": 6981
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.045180443454866e-06,
-      "loss": 0.2459,
+      "loss": 0.0997,
       "step": 6982
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.043324983764729e-06,
-      "loss": 0.2932,
+      "loss": 0.0252,
       "step": 6983
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.04146952407459e-06,
-      "loss": 0.0861,
+      "loss": 0.0751,
       "step": 6984
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.039614064384451e-06,
-      "loss": 0.2345,
+      "loss": 0.0726,
       "step": 6985
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.037758604694314e-06,
-      "loss": 0.138,
+      "loss": 0.0245,
       "step": 6986
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.035903145004175e-06,
-      "loss": 0.1379,
+      "loss": 0.1192,
       "step": 6987
     },
     {
       "epoch": 1.94,
       "learning_rate": 7.0340476853140374e-06,
-      "loss": 0.0888,
+      "loss": 0.0627,
       "step": 6988
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.0321922256238985e-06,
-      "loss": 0.3976,
+      "loss": 0.1683,
       "step": 6989
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.03033676593376e-06,
-      "loss": 0.0873,
+      "loss": 0.088,
       "step": 6990
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.028481306243622e-06,
-      "loss": 0.1408,
+      "loss": 0.1085,
       "step": 6991
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.026625846553484e-06,
-      "loss": 0.2368,
+      "loss": 0.0923,
       "step": 6992
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.024770386863345e-06,
-      "loss": 0.0896,
+      "loss": 0.1133,
       "step": 6993
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.022914927173208e-06,
-      "loss": 0.1918,
+      "loss": 0.1033,
       "step": 6994
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.021059467483069e-06,
-      "loss": 0.0913,
+      "loss": 0.0968,
       "step": 6995
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.0192040077929316e-06,
-      "loss": 0.141,
+      "loss": 0.0556,
       "step": 6996
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.017348548102793e-06,
-      "loss": 0.0932,
+      "loss": 0.1025,
       "step": 6997
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.0154930884126545e-06,
-      "loss": 0.244,
+      "loss": 0.0498,
       "step": 6998
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.013637628722516e-06,
-      "loss": 0.0935,
+      "loss": 0.0961,
       "step": 6999
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.011782169032378e-06,
-      "loss": 0.0922,
+      "loss": 0.0577,
       "step": 7000
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.009926709342239e-06,
-      "loss": 0.2444,
+      "loss": 0.0493,
       "step": 7001
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.008071249652102e-06,
-      "loss": 0.0904,
+      "loss": 0.0769,
       "step": 7002
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.006215789961963e-06,
-      "loss": 0.3348,
+      "loss": 0.0919,
       "step": 7003
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.004360330271826e-06,
-      "loss": 0.1384,
+      "loss": 0.143,
       "step": 7004
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.002504870581687e-06,
-      "loss": 0.091,
+      "loss": 0.0201,
       "step": 7005
     },
     {
       "epoch": 1.95,
       "learning_rate": 7.0006494108915494e-06,
-      "loss": 0.0896,
+      "loss": 0.1066,
       "step": 7006
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.9987939512014104e-06,
-      "loss": 0.0914,
+      "loss": 0.1008,
       "step": 7007
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.996938491511272e-06,
-      "loss": 0.1922,
+      "loss": 0.1413,
       "step": 7008
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.995083031821134e-06,
-      "loss": 0.0893,
+      "loss": 0.0907,
       "step": 7009
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.993227572130996e-06,
-      "loss": 0.0878,
+      "loss": 0.081,
       "step": 7010
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.991372112440857e-06,
-      "loss": 0.0391,
+      "loss": 0.0276,
       "step": 7011
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.98951665275072e-06,
-      "loss": 0.2386,
+      "loss": 0.0826,
       "step": 7012
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.987661193060581e-06,
-      "loss": 0.1415,
+      "loss": 0.0901,
       "step": 7013
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.9858057333704435e-06,
-      "loss": 0.1888,
+      "loss": 0.1127,
       "step": 7014
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.9839502736803046e-06,
-      "loss": 0.2382,
+      "loss": 0.0168,
       "step": 7015
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.9820948139901664e-06,
-      "loss": 0.2982,
+      "loss": 0.0211,
       "step": 7016
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.980239354300028e-06,
-      "loss": 0.0374,
+      "loss": 0.0594,
       "step": 7017
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.97838389460989e-06,
-      "loss": 0.1885,
+      "loss": 0.0548,
       "step": 7018
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.976528434919751e-06,
-      "loss": 0.0892,
+      "loss": 0.0574,
       "step": 7019
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.974672975229614e-06,
-      "loss": 0.089,
+      "loss": 0.0559,
       "step": 7020
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.972817515539475e-06,
-      "loss": 0.1396,
+      "loss": 0.0901,
       "step": 7021
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.970962055849338e-06,
-      "loss": 0.188,
+      "loss": 0.0151,
       "step": 7022
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.969106596159199e-06,
-      "loss": 0.1865,
+      "loss": 0.0575,
       "step": 7023
     },
     {
       "epoch": 1.95,
       "learning_rate": 6.9672511364690606e-06,
-      "loss": 0.1434,
+      "loss": 0.0207,
       "step": 7024
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.9653956767789224e-06,
-      "loss": 0.0362,
+      "loss": 0.0165,
       "step": 7025
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.963540217088784e-06,
-      "loss": 0.139,
+      "loss": 0.0179,
       "step": 7026
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.961684757398645e-06,
-      "loss": 0.1422,
+      "loss": 0.1608,
       "step": 7027
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.959829297708508e-06,
-      "loss": 0.036,
+      "loss": 0.021,
       "step": 7028
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.957973838018369e-06,
-      "loss": 0.1398,
+      "loss": 0.0135,
       "step": 7029
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.956118378328232e-06,
-      "loss": 0.1351,
+      "loss": 0.0901,
       "step": 7030
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.954262918638093e-06,
-      "loss": 0.1863,
+      "loss": 0.0279,
       "step": 7031
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.9524074589479555e-06,
-      "loss": 0.1917,
+      "loss": 0.0393,
       "step": 7032
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.9505519992578165e-06,
-      "loss": 0.0345,
+      "loss": 0.0387,
       "step": 7033
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.948696539567678e-06,
-      "loss": 0.089,
+      "loss": 0.0161,
       "step": 7034
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.94684107987754e-06,
-      "loss": 0.137,
+      "loss": 0.0942,
       "step": 7035
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.944985620187402e-06,
-      "loss": 0.0872,
+      "loss": 0.0353,
       "step": 7036
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.943130160497263e-06,
-      "loss": 0.1388,
+      "loss": 0.0119,
       "step": 7037
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.941274700807126e-06,
-      "loss": 0.1933,
+      "loss": 0.0463,
       "step": 7038
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.939419241116987e-06,
-      "loss": 0.0839,
+      "loss": 0.1139,
       "step": 7039
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.93756378142685e-06,
-      "loss": 0.0869,
+      "loss": 0.0631,
       "step": 7040
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.935708321736711e-06,
-      "loss": 0.1395,
+      "loss": 0.089,
       "step": 7041
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.9338528620465725e-06,
-      "loss": 0.1927,
+      "loss": 0.013,
       "step": 7042
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.931997402356434e-06,
-      "loss": 0.0869,
+      "loss": 0.0548,
       "step": 7043
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.930141942666296e-06,
-      "loss": 0.1327,
+      "loss": 0.0549,
       "step": 7044
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.928286482976157e-06,
-      "loss": 0.1386,
+      "loss": 0.0471,
       "step": 7045
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.92643102328602e-06,
-      "loss": 0.0835,
+      "loss": 0.1921,
       "step": 7046
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.924575563595881e-06,
-      "loss": 0.1914,
+      "loss": 0.0424,
       "step": 7047
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.922720103905744e-06,
-      "loss": 0.1372,
+      "loss": 0.109,
       "step": 7048
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.920864644215605e-06,
-      "loss": 0.252,
+      "loss": 0.0116,
       "step": 7049
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.919009184525467e-06,
-      "loss": 0.3024,
+      "loss": 0.0327,
       "step": 7050
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.9171537248353285e-06,
-      "loss": 0.1419,
+      "loss": 0.0063,
       "step": 7051
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.91529826514519e-06,
-      "loss": 0.0839,
+      "loss": 0.0862,
       "step": 7052
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.913442805455051e-06,
-      "loss": 0.1381,
+      "loss": 0.1677,
       "step": 7053
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.911587345764914e-06,
-      "loss": 0.0842,
+      "loss": 0.0173,
       "step": 7054
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.909731886074775e-06,
-      "loss": 0.1376,
+      "loss": 0.1614,
       "step": 7055
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.907876426384638e-06,
-      "loss": 0.193,
+      "loss": 0.0502,
       "step": 7056
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.906020966694499e-06,
-      "loss": 0.1929,
+      "loss": 0.1138,
       "step": 7057
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.904165507004361e-06,
-      "loss": 0.1414,
+      "loss": 0.0094,
       "step": 7058
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.902310047314223e-06,
-      "loss": 0.0824,
+      "loss": 0.0527,
       "step": 7059
     },
     {
       "epoch": 1.96,
       "learning_rate": 6.9004545876240845e-06,
-      "loss": 0.0824,
+      "loss": 0.1079,
       "step": 7060
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.8985991279339455e-06,
-      "loss": 0.191,
+      "loss": 0.115,
       "step": 7061
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.896743668243808e-06,
-      "loss": 0.1965,
+      "loss": 0.0164,
       "step": 7062
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.894888208553669e-06,
-      "loss": 0.0323,
+      "loss": 0.0123,
       "step": 7063
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.893032748863532e-06,
-      "loss": 0.2467,
+      "loss": 0.0343,
       "step": 7064
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.891177289173393e-06,
-      "loss": 0.0875,
+      "loss": 0.0839,
       "step": 7065
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.889321829483256e-06,
-      "loss": 0.1425,
+      "loss": 0.1177,
       "step": 7066
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.887466369793117e-06,
-      "loss": 0.0868,
+      "loss": 0.0211,
       "step": 7067
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.885610910102979e-06,
-      "loss": 0.0881,
+      "loss": 0.0474,
       "step": 7068
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.8837554504128405e-06,
-      "loss": 0.1363,
+      "loss": 0.1186,
       "step": 7069
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.881899990722702e-06,
-      "loss": 0.086,
+      "loss": 0.0438,
       "step": 7070
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.880044531032563e-06,
-      "loss": 0.2963,
+      "loss": 0.1342,
       "step": 7071
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.878189071342426e-06,
-      "loss": 0.0849,
+      "loss": 0.1293,
       "step": 7072
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.876333611652287e-06,
-      "loss": 0.1386,
+      "loss": 0.0259,
       "step": 7073
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.87447815196215e-06,
-      "loss": 0.0866,
+      "loss": 0.0809,
       "step": 7074
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.872622692272011e-06,
-      "loss": 0.0842,
+      "loss": 0.1486,
       "step": 7075
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.870767232581873e-06,
-      "loss": 0.2968,
+      "loss": 0.153,
       "step": 7076
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.868911772891735e-06,
-      "loss": 0.1371,
+      "loss": 0.0891,
       "step": 7077
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.8670563132015965e-06,
-      "loss": 0.2461,
+      "loss": 0.1077,
       "step": 7078
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.8652008535114575e-06,
-      "loss": 0.1955,
+      "loss": 0.099,
       "step": 7079
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.86334539382132e-06,
-      "loss": 0.1397,
+      "loss": 0.0759,
       "step": 7080
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.861489934131181e-06,
-      "loss": 0.1386,
+      "loss": 0.0466,
       "step": 7081
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.859634474441044e-06,
-      "loss": 0.3023,
+      "loss": 0.067,
       "step": 7082
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.857779014750905e-06,
-      "loss": 0.2474,
+      "loss": 0.0697,
       "step": 7083
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.855923555060767e-06,
-      "loss": 0.1921,
+      "loss": 0.0676,
       "step": 7084
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.854068095370629e-06,
-      "loss": 0.1949,
+      "loss": 0.0625,
       "step": 7085
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.852212635680491e-06,
-      "loss": 0.2965,
+      "loss": 0.0808,
       "step": 7086
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.850357175990352e-06,
-      "loss": 0.1364,
+      "loss": 0.015,
       "step": 7087
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.848501716300214e-06,
-      "loss": 0.1427,
+      "loss": 0.028,
       "step": 7088
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.846646256610075e-06,
-      "loss": 0.1392,
+      "loss": 0.0786,
       "step": 7089
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.844790796919938e-06,
-      "loss": 0.1378,
+      "loss": 0.1014,
       "step": 7090
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.842935337229799e-06,
-      "loss": 0.1423,
+      "loss": 0.0907,
       "step": 7091
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.841079877539662e-06,
-      "loss": 0.1906,
+      "loss": 0.0194,
       "step": 7092
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.839224417849523e-06,
-      "loss": 0.1918,
+      "loss": 0.0537,
       "step": 7093
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.837368958159385e-06,
-      "loss": 0.0912,
+      "loss": 0.1664,
       "step": 7094
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.835513498469247e-06,
-      "loss": 0.0895,
+      "loss": 0.0156,
       "step": 7095
     },
     {
       "epoch": 1.97,
       "learning_rate": 6.8336580387791085e-06,
-      "loss": 0.0376,
+      "loss": 0.0842,
       "step": 7096
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.8318025790889695e-06,
-      "loss": 0.1367,
+      "loss": 0.09,
       "step": 7097
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.829947119398832e-06,
-      "loss": 0.1423,
+      "loss": 0.022,
       "step": 7098
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.828091659708693e-06,
-      "loss": 0.1353,
+      "loss": 0.0287,
       "step": 7099
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.826236200018556e-06,
-      "loss": 0.2396,
+      "loss": 0.0581,
       "step": 7100
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.824380740328417e-06,
-      "loss": 0.0875,
+      "loss": 0.0156,
       "step": 7101
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.822525280638279e-06,
-      "loss": 0.1912,
+      "loss": 0.248,
       "step": 7102
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.820669820948141e-06,
-      "loss": 0.0367,
+      "loss": 0.0455,
       "step": 7103
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.818814361258003e-06,
-      "loss": 0.1357,
+      "loss": 0.0964,
       "step": 7104
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.816958901567864e-06,
-      "loss": 0.1393,
+      "loss": 0.0659,
       "step": 7105
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.815103441877726e-06,
-      "loss": 0.1382,
+      "loss": 0.0875,
       "step": 7106
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.813247982187587e-06,
-      "loss": 0.0888,
+      "loss": 0.0595,
       "step": 7107
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.81139252249745e-06,
-      "loss": 0.0358,
+      "loss": 0.0738,
       "step": 7108
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.809537062807311e-06,
-      "loss": 0.1355,
+      "loss": 0.0204,
       "step": 7109
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.807681603117173e-06,
-      "loss": 0.0341,
+      "loss": 0.0607,
       "step": 7110
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.805826143427035e-06,
-      "loss": 0.0345,
+      "loss": 0.1002,
       "step": 7111
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.803970683736897e-06,
-      "loss": 0.1917,
+      "loss": 0.0874,
       "step": 7112
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.802115224046758e-06,
-      "loss": 0.0322,
+      "loss": 0.0178,
       "step": 7113
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.8002597643566204e-06,
-      "loss": 0.0332,
+      "loss": 0.024,
       "step": 7114
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.7984043046664815e-06,
-      "loss": 0.0313,
+      "loss": 0.0444,
       "step": 7115
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.796548844976344e-06,
-      "loss": 0.1385,
+      "loss": 0.0293,
       "step": 7116
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.794693385286205e-06,
-      "loss": 0.2455,
+      "loss": 0.0492,
       "step": 7117
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.792837925596067e-06,
-      "loss": 0.4086,
+      "loss": 0.0482,
       "step": 7118
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.790982465905929e-06,
-      "loss": 0.1899,
+      "loss": 0.0156,
       "step": 7119
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.789127006215791e-06,
-      "loss": 0.1382,
+      "loss": 0.0813,
       "step": 7120
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.787271546525652e-06,
-      "loss": 0.0848,
+      "loss": 0.0633,
       "step": 7121
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.7854160868355146e-06,
-      "loss": 0.138,
+      "loss": 0.0123,
       "step": 7122
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.783560627145376e-06,
-      "loss": 0.1937,
+      "loss": 0.0959,
       "step": 7123
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.781705167455238e-06,
-      "loss": 0.0319,
+      "loss": 0.0108,
       "step": 7124
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.779849707765099e-06,
-      "loss": 0.0815,
+      "loss": 0.0108,
       "step": 7125
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.777994248074962e-06,
-      "loss": 0.2465,
+      "loss": 0.1496,
       "step": 7126
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.776138788384823e-06,
-      "loss": 0.1923,
+      "loss": 0.0136,
       "step": 7127
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.774283328694685e-06,
-      "loss": 0.192,
+      "loss": 0.0103,
       "step": 7128
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.772427869004547e-06,
-      "loss": 0.03,
+      "loss": 0.0085,
       "step": 7129
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.770572409314409e-06,
-      "loss": 0.201,
+      "loss": 0.1029,
       "step": 7130
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.76871694962427e-06,
-      "loss": 0.1352,
+      "loss": 0.0114,
       "step": 7131
     },
     {
       "epoch": 1.98,
       "learning_rate": 6.766861489934132e-06,
-      "loss": 0.1404,
+      "loss": 0.0942,
       "step": 7132
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.7650060302439934e-06,
-      "loss": 0.0825,
+      "loss": 0.0873,
       "step": 7133
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.763150570553856e-06,
-      "loss": 0.1935,
+      "loss": 0.0081,
       "step": 7134
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.761295110863717e-06,
-      "loss": 0.1873,
+      "loss": 0.0122,
       "step": 7135
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.759439651173578e-06,
-      "loss": 0.2456,
+      "loss": 0.139,
       "step": 7136
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.757584191483441e-06,
-      "loss": 0.0306,
+      "loss": 0.0875,
       "step": 7137
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.755728731793302e-06,
-      "loss": 0.0848,
+      "loss": 0.1437,
       "step": 7138
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.753873272103164e-06,
-      "loss": 0.1936,
+      "loss": 0.1574,
       "step": 7139
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.752017812413026e-06,
-      "loss": 0.2473,
+      "loss": 0.0491,
       "step": 7140
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.7501623527228876e-06,
-      "loss": 0.0849,
+      "loss": 0.0123,
       "step": 7141
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.748306893032749e-06,
-      "loss": 0.1398,
+      "loss": 0.1197,
       "step": 7142
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.746451433342611e-06,
-      "loss": 0.0309,
+      "loss": 0.012,
       "step": 7143
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.744595973652472e-06,
-      "loss": 0.0849,
+      "loss": 0.0656,
       "step": 7144
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.742740513962335e-06,
-      "loss": 0.0859,
+      "loss": 0.0895,
       "step": 7145
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.740885054272196e-06,
-      "loss": 0.0316,
+      "loss": 0.1431,
       "step": 7146
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.739029594582058e-06,
-      "loss": 0.1354,
+      "loss": 0.0719,
       "step": 7147
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.73717413489192e-06,
-      "loss": 0.2925,
+      "loss": 0.0283,
       "step": 7148
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.735318675201782e-06,
-      "loss": 0.1428,
+      "loss": 0.0534,
       "step": 7149
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.733463215511643e-06,
-      "loss": 0.1388,
+      "loss": 0.0196,
       "step": 7150
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.731607755821505e-06,
-      "loss": 0.1349,
+      "loss": 0.0631,
       "step": 7151
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.7297522961313664e-06,
-      "loss": 0.1378,
+      "loss": 0.0494,
       "step": 7152
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.727896836441229e-06,
-      "loss": 0.2434,
+      "loss": 0.0952,
       "step": 7153
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.72604137675109e-06,
-      "loss": 0.1401,
+      "loss": 0.0608,
       "step": 7154
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.724185917060953e-06,
-      "loss": 0.0859,
+      "loss": 0.0131,
       "step": 7155
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.722330457370814e-06,
-      "loss": 0.0839,
+      "loss": 0.0661,
       "step": 7156
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.720474997680676e-06,
-      "loss": 0.0319,
+      "loss": 0.0504,
       "step": 7157
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.718619537990537e-06,
-      "loss": 0.1363,
+      "loss": 0.0623,
       "step": 7158
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.7167640783003995e-06,
-      "loss": 0.2515,
+      "loss": 0.3217,
       "step": 7159
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.7149086186102606e-06,
-      "loss": 0.1405,
+      "loss": 0.0714,
       "step": 7160
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.713053158920123e-06,
-      "loss": 0.1959,
+      "loss": 0.0645,
       "step": 7161
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.711197699229984e-06,
-      "loss": 0.3524,
+      "loss": 0.1013,
       "step": 7162
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.709342239539847e-06,
-      "loss": 0.0828,
+      "loss": 0.0907,
       "step": 7163
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.707486779849708e-06,
-      "loss": 0.1381,
+      "loss": 0.1682,
       "step": 7164
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.70563132015957e-06,
-      "loss": 0.1931,
+      "loss": 0.069,
       "step": 7165
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.703775860469432e-06,
-      "loss": 0.1949,
+      "loss": 0.14,
       "step": 7166
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.701920400779294e-06,
-      "loss": 0.1404,
+      "loss": 0.0202,
       "step": 7167
     },
     {
       "epoch": 1.99,
       "learning_rate": 6.700064941089155e-06,
-      "loss": 0.2447,
+      "loss": 0.0445,
       "step": 7168
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.698209481399017e-06,
-      "loss": 0.1392,
+      "loss": 0.0964,
       "step": 7169
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.696354021708878e-06,
-      "loss": 0.0324,
+      "loss": 0.0954,
       "step": 7170
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.694498562018741e-06,
-      "loss": 0.1465,
+      "loss": 0.0327,
       "step": 7171
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.692643102328602e-06,
-      "loss": 0.1394,
+      "loss": 0.1476,
       "step": 7172
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.690787642638464e-06,
-      "loss": 0.1879,
+      "loss": 0.0252,
       "step": 7173
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.688932182948326e-06,
-      "loss": 0.1387,
+      "loss": 0.0221,
       "step": 7174
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.687076723258188e-06,
-      "loss": 0.0846,
+      "loss": 0.0572,
       "step": 7175
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.685221263568049e-06,
-      "loss": 0.0864,
+      "loss": 0.0571,
       "step": 7176
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.6833658038779115e-06,
-      "loss": 0.191,
+      "loss": 0.0822,
       "step": 7177
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.6815103441877725e-06,
-      "loss": 0.1416,
+      "loss": 0.1058,
       "step": 7178
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.679654884497635e-06,
-      "loss": 0.2421,
+      "loss": 0.0241,
       "step": 7179
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.677799424807496e-06,
-      "loss": 0.1379,
+      "loss": 0.0528,
       "step": 7180
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.675943965117358e-06,
-      "loss": 0.086,
+      "loss": 0.0192,
       "step": 7181
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.67408850542722e-06,
-      "loss": 0.2972,
+      "loss": 0.0279,
       "step": 7182
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.672233045737082e-06,
-      "loss": 0.1902,
+      "loss": 0.0194,
       "step": 7183
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.670377586046943e-06,
-      "loss": 0.0876,
+      "loss": 0.0612,
       "step": 7184
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.668522126356806e-06,
-      "loss": 0.0339,
+      "loss": 0.1429,
       "step": 7185
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.0341,
+      "loss": 0.151,
       "step": 7186
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.664811206976529e-06,
-      "loss": 0.0333,
+      "loss": 0.1564,
       "step": 7187
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.66295574728639e-06,
-      "loss": 0.1916,
+      "loss": 0.0286,
       "step": 7188
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.661100287596253e-06,
-      "loss": 0.1392,
+      "loss": 0.0162,
       "step": 7189
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.659244827906114e-06,
-      "loss": 0.1376,
+      "loss": 0.0838,
       "step": 7190
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.657389368215976e-06,
-      "loss": 0.0343,
+      "loss": 0.0414,
       "step": 7191
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.655533908525838e-06,
-      "loss": 0.0337,
+      "loss": 0.0645,
       "step": 7192
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.6536784488357e-06,
-      "loss": 0.1929,
+      "loss": 0.0971,
       "step": 7193
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.651822989145561e-06,
-      "loss": 0.1352,
+      "loss": 0.0145,
       "step": 7194
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.6499675294554235e-06,
-      "loss": 0.1414,
+      "loss": 0.09,
       "step": 7195
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.6481120697652845e-06,
-      "loss": 0.085,
+      "loss": 0.0506,
       "step": 7196
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.646256610075147e-06,
-      "loss": 0.0855,
+      "loss": 0.1503,
       "step": 7197
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.644401150385008e-06,
-      "loss": 0.0853,
+      "loss": 0.063,
       "step": 7198
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.64254569069487e-06,
-      "loss": 0.0873,
+      "loss": 0.0129,
       "step": 7199
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.640690231004732e-06,
-      "loss": 0.2509,
+      "loss": 0.0159,
       "step": 7200
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.638834771314594e-06,
-      "loss": 0.0309,
+      "loss": 0.066,
       "step": 7201
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.636979311624455e-06,
-      "loss": 0.1981,
+      "loss": 0.0069,
       "step": 7202
     },
     {
       "epoch": 2.0,
       "learning_rate": 6.635123851934318e-06,
-      "loss": 0.3026,
+      "loss": 0.0462,
       "step": 7203
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.633268392244179e-06,
-      "loss": 0.1398,
+      "loss": 0.0882,
       "step": 7204
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.631412932554041e-06,
-      "loss": 0.0853,
+      "loss": 0.0435,
       "step": 7205
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.629557472863902e-06,
-      "loss": 0.0869,
+      "loss": 0.105,
       "step": 7206
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.627702013173764e-06,
-      "loss": 0.0826,
+      "loss": 0.0135,
       "step": 7207
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.625846553483626e-06,
-      "loss": 0.2457,
+      "loss": 0.0572,
       "step": 7208
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.623991093793488e-06,
-      "loss": 0.1396,
+      "loss": 0.0181,
       "step": 7209
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.622135634103349e-06,
-      "loss": 0.0846,
+      "loss": 0.0678,
       "step": 7210
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.620280174413212e-06,
-      "loss": 0.3025,
+      "loss": 0.1533,
       "step": 7211
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.618424714723073e-06,
-      "loss": 0.249,
+      "loss": 0.1748,
       "step": 7212
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.6165692550329355e-06,
-      "loss": 0.1359,
+      "loss": 0.0478,
       "step": 7213
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.6147137953427965e-06,
-      "loss": 0.3055,
+      "loss": 0.0213,
       "step": 7214
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.612858335652659e-06,
-      "loss": 0.0317,
+      "loss": 0.0137,
       "step": 7215
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.61100287596252e-06,
-      "loss": 0.1906,
+      "loss": 0.0596,
       "step": 7216
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.609147416272382e-06,
-      "loss": 0.1844,
+      "loss": 0.1735,
       "step": 7217
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.607291956582243e-06,
-      "loss": 0.2443,
+      "loss": 0.0988,
       "step": 7218
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.605436496892106e-06,
-      "loss": 0.1427,
+      "loss": 0.0604,
       "step": 7219
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.603581037201967e-06,
-      "loss": 0.0844,
+      "loss": 0.0184,
       "step": 7220
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.6017255775118296e-06,
-      "loss": 0.1399,
+      "loss": 0.0484,
       "step": 7221
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.599870117821691e-06,
-      "loss": 0.1397,
+      "loss": 0.2232,
       "step": 7222
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.598014658131553e-06,
-      "loss": 0.0861,
+      "loss": 0.0137,
       "step": 7223
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.596159198441414e-06,
-      "loss": 0.1437,
+      "loss": 0.039,
       "step": 7224
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.594303738751276e-06,
-      "loss": 0.1364,
+      "loss": 0.0513,
       "step": 7225
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.592448279061138e-06,
-      "loss": 0.0868,
+      "loss": 0.1222,
       "step": 7226
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.590592819371e-06,
-      "loss": 0.0332,
+      "loss": 0.206,
       "step": 7227
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.588737359680861e-06,
-      "loss": 0.3507,
+      "loss": 0.0953,
       "step": 7228
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.586881899990724e-06,
-      "loss": 0.2422,
+      "loss": 0.0983,
       "step": 7229
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.585026440300585e-06,
-      "loss": 0.1377,
+      "loss": 0.0637,
       "step": 7230
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.5831709806104474e-06,
-      "loss": 0.1903,
+      "loss": 0.0699,
       "step": 7231
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.5813155209203085e-06,
-      "loss": 0.2982,
+      "loss": 0.0705,
       "step": 7232
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.57946006123017e-06,
-      "loss": 0.2935,
+      "loss": 0.0692,
       "step": 7233
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.577604601540032e-06,
-      "loss": 0.1442,
+      "loss": 0.0223,
       "step": 7234
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.575749141849894e-06,
-      "loss": 0.0877,
+      "loss": 0.0597,
       "step": 7235
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.573893682159755e-06,
-      "loss": 0.1874,
+      "loss": 0.0493,
       "step": 7236
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.572038222469618e-06,
-      "loss": 0.0358,
+      "loss": 0.0914,
       "step": 7237
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.570182762779479e-06,
-      "loss": 0.238,
+      "loss": 0.075,
       "step": 7238
     },
     {
       "epoch": 2.01,
       "learning_rate": 6.5683273030893416e-06,
-      "loss": 0.0883,
+      "loss": 0.0495,
       "step": 7239
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.5664718433992026e-06,
-      "loss": 0.2446,
+      "loss": 0.0594,
       "step": 7240
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.5646163837090644e-06,
-      "loss": 0.1388,
+      "loss": 0.1017,
       "step": 7241
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.562760924018926e-06,
-      "loss": 0.0363,
+      "loss": 0.0515,
       "step": 7242
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.560905464328788e-06,
-      "loss": 0.0898,
+      "loss": 0.1516,
       "step": 7243
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.559050004638649e-06,
-      "loss": 0.2372,
+      "loss": 0.0832,
       "step": 7244
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.557194544948512e-06,
-      "loss": 0.0876,
+      "loss": 0.0471,
       "step": 7245
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.555339085258373e-06,
-      "loss": 0.0388,
+      "loss": 0.1114,
       "step": 7246
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.553483625568236e-06,
-      "loss": 0.1888,
+      "loss": 0.1062,
       "step": 7247
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.551628165878097e-06,
-      "loss": 0.2444,
+      "loss": 0.1047,
       "step": 7248
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.549772706187959e-06,
-      "loss": 0.1411,
+      "loss": 0.1935,
       "step": 7249
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.5479172464978204e-06,
-      "loss": 0.191,
+      "loss": 0.016,
       "step": 7250
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.546061786807682e-06,
-      "loss": 0.1957,
+      "loss": 0.0756,
       "step": 7251
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.544206327117544e-06,
-      "loss": 0.2448,
+      "loss": 0.0286,
       "step": 7252
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.542350867427406e-06,
-      "loss": 0.1389,
+      "loss": 0.0182,
       "step": 7253
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.540495407737267e-06,
-      "loss": 0.192,
+      "loss": 0.1106,
       "step": 7254
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.53863994804713e-06,
-      "loss": 0.0882,
+      "loss": 0.0469,
       "step": 7255
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.536784488356991e-06,
-      "loss": 0.1872,
+      "loss": 0.0223,
       "step": 7256
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.5349290286668535e-06,
-      "loss": 0.0393,
+      "loss": 0.0182,
       "step": 7257
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.5330735689767146e-06,
-      "loss": 0.0384,
+      "loss": 0.1813,
       "step": 7258
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.531218109286576e-06,
-      "loss": 0.0375,
+      "loss": 0.2278,
       "step": 7259
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.529362649596438e-06,
-      "loss": 0.2368,
+      "loss": 0.1227,
       "step": 7260
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.5275071899063e-06,
-      "loss": 0.1408,
+      "loss": 0.0507,
       "step": 7261
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.525651730216161e-06,
-      "loss": 0.1395,
+      "loss": 0.0485,
       "step": 7262
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.523796270526024e-06,
-      "loss": 0.09,
+      "loss": 0.0185,
       "step": 7263
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.521940810835885e-06,
-      "loss": 0.1416,
+      "loss": 0.1039,
       "step": 7264
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.520085351145748e-06,
-      "loss": 0.143,
+      "loss": 0.0253,
       "step": 7265
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.518229891455609e-06,
-      "loss": 0.0368,
+      "loss": 0.1284,
       "step": 7266
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.5163744317654705e-06,
-      "loss": 0.0892,
+      "loss": 0.0559,
       "step": 7267
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.514518972075332e-06,
-      "loss": 0.0349,
+      "loss": 0.0471,
       "step": 7268
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.512663512385194e-06,
-      "loss": 0.1396,
+      "loss": 0.0488,
       "step": 7269
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.510808052695055e-06,
-      "loss": 0.0878,
+      "loss": 0.0659,
       "step": 7270
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.508952593004918e-06,
-      "loss": 0.1908,
+      "loss": 0.1349,
       "step": 7271
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.507097133314779e-06,
-      "loss": 0.1372,
+      "loss": 0.1472,
       "step": 7272
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.505241673624642e-06,
-      "loss": 0.0842,
+      "loss": 0.0207,
       "step": 7273
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.503386213934503e-06,
-      "loss": 0.1952,
+      "loss": 0.0514,
       "step": 7274
     },
     {
       "epoch": 2.02,
       "learning_rate": 6.5015307542443655e-06,
-      "loss": 0.3027,
+      "loss": 0.0548,
       "step": 7275
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.4996752945542265e-06,
-      "loss": 0.1946,
+      "loss": 0.1043,
       "step": 7276
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.497819834864088e-06,
-      "loss": 0.0327,
+      "loss": 0.0275,
       "step": 7277
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.495964375173949e-06,
-      "loss": 0.2481,
+      "loss": 0.1328,
       "step": 7278
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.494108915483812e-06,
-      "loss": 0.0831,
+      "loss": 0.0161,
       "step": 7279
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.492253455793673e-06,
-      "loss": 0.0857,
+      "loss": 0.0764,
       "step": 7280
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.490397996103536e-06,
-      "loss": 0.0819,
+      "loss": 0.0784,
       "step": 7281
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.488542536413397e-06,
-      "loss": 0.1969,
+      "loss": 0.0167,
       "step": 7282
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.48668707672326e-06,
-      "loss": 0.0319,
+      "loss": 0.0563,
       "step": 7283
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.484831617033121e-06,
-      "loss": 0.0854,
+      "loss": 0.1368,
       "step": 7284
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.4829761573429825e-06,
-      "loss": 0.1424,
+      "loss": 0.0375,
       "step": 7285
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.481120697652844e-06,
-      "loss": 0.3022,
+      "loss": 0.1058,
       "step": 7286
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.479265237962706e-06,
-      "loss": 0.1379,
+      "loss": 0.1601,
       "step": 7287
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.477409778272567e-06,
-      "loss": 0.0846,
+      "loss": 0.0237,
       "step": 7288
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.475554318582429e-06,
-      "loss": 0.0867,
+      "loss": 0.0482,
       "step": 7289
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.473698858892291e-06,
-      "loss": 0.1967,
+      "loss": 0.0725,
       "step": 7290
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.471843399202152e-06,
-      "loss": 0.1856,
+      "loss": 0.047,
       "step": 7291
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.469987939512015e-06,
-      "loss": 0.1958,
+      "loss": 0.0742,
       "step": 7292
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.468132479821876e-06,
-      "loss": 0.0304,
+      "loss": 0.1652,
       "step": 7293
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.4662770201317385e-06,
-      "loss": 0.0314,
+      "loss": 0.0103,
       "step": 7294
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.4644215604415995e-06,
-      "loss": 0.0306,
+      "loss": 0.1296,
       "step": 7295
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.462566100751461e-06,
-      "loss": 0.1962,
+      "loss": 0.028,
       "step": 7296
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.460710641061323e-06,
-      "loss": 0.2444,
+      "loss": 0.0721,
       "step": 7297
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.458855181371185e-06,
-      "loss": 0.0292,
+      "loss": 0.0191,
       "step": 7298
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.456999721681046e-06,
-      "loss": 0.0309,
+      "loss": 0.01,
       "step": 7299
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.455144261990909e-06,
-      "loss": 0.2512,
+      "loss": 0.0331,
       "step": 7300
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.45328880230077e-06,
-      "loss": 0.0859,
+      "loss": 0.0147,
       "step": 7301
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.451433342610633e-06,
-      "loss": 0.1929,
+      "loss": 0.1454,
       "step": 7302
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.449577882920494e-06,
-      "loss": 0.1357,
+      "loss": 0.0077,
       "step": 7303
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.4477224232303555e-06,
-      "loss": 0.1376,
+      "loss": 0.0676,
       "step": 7304
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.445866963540217e-06,
-      "loss": 0.0872,
+      "loss": 0.1161,
       "step": 7305
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.444011503850079e-06,
-      "loss": 0.0842,
+      "loss": 0.3127,
       "step": 7306
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.44215604415994e-06,
-      "loss": 0.1376,
+      "loss": 0.0211,
       "step": 7307
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.440300584469803e-06,
-      "loss": 0.0305,
+      "loss": 0.0865,
       "step": 7308
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.438445124779664e-06,
-      "loss": 0.0295,
+      "loss": 0.0935,
       "step": 7309
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.436589665089527e-06,
-      "loss": 0.1383,
+      "loss": 0.1692,
       "step": 7310
     },
     {
       "epoch": 2.03,
       "learning_rate": 6.434734205399388e-06,
-      "loss": 0.1973,
+      "loss": 0.0856,
       "step": 7311
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.4328787457092505e-06,
-      "loss": 0.2473,
+      "loss": 0.1439,
       "step": 7312
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.4310232860191115e-06,
-      "loss": 0.1972,
+      "loss": 0.0278,
       "step": 7313
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.429167826328973e-06,
-      "loss": 0.1945,
+      "loss": 0.1135,
       "step": 7314
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.427312366638835e-06,
-      "loss": 0.1384,
+      "loss": 0.1859,
       "step": 7315
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.425456906948697e-06,
-      "loss": 0.0821,
+      "loss": 0.0661,
       "step": 7316
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.423601447258558e-06,
-      "loss": 0.083,
+      "loss": 0.0969,
       "step": 7317
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.421745987568421e-06,
-      "loss": 0.1389,
+      "loss": 0.1079,
       "step": 7318
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.419890527878282e-06,
-      "loss": 0.2453,
+      "loss": 0.0339,
       "step": 7319
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.418035068188145e-06,
-      "loss": 0.137,
+      "loss": 0.0224,
       "step": 7320
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.416179608498006e-06,
-      "loss": 0.0295,
+      "loss": 0.0559,
       "step": 7321
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.4143241488078675e-06,
-      "loss": 0.1962,
+      "loss": 0.1072,
       "step": 7322
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.412468689117729e-06,
-      "loss": 0.1387,
+      "loss": 0.1114,
       "step": 7323
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.410613229427591e-06,
-      "loss": 0.4162,
+      "loss": 0.0308,
       "step": 7324
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.408757769737452e-06,
-      "loss": 0.1412,
+      "loss": 0.0829,
       "step": 7325
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.406902310047315e-06,
-      "loss": 0.1355,
+      "loss": 0.0702,
       "step": 7326
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.405046850357176e-06,
-      "loss": 0.1952,
+      "loss": 0.0491,
       "step": 7327
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.403191390667039e-06,
-      "loss": 0.3451,
+      "loss": 0.0501,
       "step": 7328
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.4013359309769e-06,
-      "loss": 0.2489,
+      "loss": 0.0375,
       "step": 7329
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.399480471286762e-06,
-      "loss": 0.0311,
+      "loss": 0.0946,
       "step": 7330
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.3976250115966235e-06,
-      "loss": 0.1356,
+      "loss": 0.2773,
       "step": 7331
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.395769551906485e-06,
-      "loss": 0.0855,
+      "loss": 0.096,
       "step": 7332
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.393914092216346e-06,
-      "loss": 0.1959,
+      "loss": 0.0755,
       "step": 7333
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.392058632526209e-06,
-      "loss": 0.0326,
+      "loss": 0.1286,
       "step": 7334
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.39020317283607e-06,
-      "loss": 0.1343,
+      "loss": 0.0078,
       "step": 7335
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.388347713145933e-06,
-      "loss": 0.2419,
+      "loss": 0.0128,
       "step": 7336
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.386492253455794e-06,
-      "loss": 0.139,
+      "loss": 0.1265,
       "step": 7337
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.384636793765656e-06,
-      "loss": 0.1398,
+      "loss": 0.0215,
       "step": 7338
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.382781334075518e-06,
-      "loss": 0.1336,
+      "loss": 0.0224,
       "step": 7339
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.3809258743853795e-06,
-      "loss": 0.2965,
+      "loss": 0.109,
       "step": 7340
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.3790704146952405e-06,
-      "loss": 0.1892,
+      "loss": 0.2325,
       "step": 7341
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.377214955005103e-06,
-      "loss": 0.0866,
+      "loss": 0.0186,
       "step": 7342
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.375359495314964e-06,
-      "loss": 0.2449,
+      "loss": 0.0155,
       "step": 7343
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.373504035624827e-06,
-      "loss": 0.0347,
+      "loss": 0.146,
       "step": 7344
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.371648575934688e-06,
-      "loss": 0.2985,
+      "loss": 0.1458,
       "step": 7345
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.369793116244551e-06,
-      "loss": 0.0854,
+      "loss": 0.0516,
       "step": 7346
     },
     {
       "epoch": 2.04,
       "learning_rate": 6.367937656554412e-06,
-      "loss": 0.2462,
+      "loss": 0.1957,
       "step": 7347
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.366082196864274e-06,
-      "loss": 0.0876,
+      "loss": 0.0736,
       "step": 7348
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.3642267371741355e-06,
-      "loss": 0.0876,
+      "loss": 0.0609,
       "step": 7349
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.362371277483997e-06,
-      "loss": 0.1955,
+      "loss": 0.0206,
       "step": 7350
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.360515817793858e-06,
-      "loss": 0.0874,
+      "loss": 0.055,
       "step": 7351
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.358660358103721e-06,
-      "loss": 0.0875,
+      "loss": 0.1092,
       "step": 7352
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.356804898413582e-06,
-      "loss": 0.0893,
+      "loss": 0.0247,
       "step": 7353
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.354949438723445e-06,
-      "loss": 0.0362,
+      "loss": 0.0967,
       "step": 7354
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.353093979033306e-06,
-      "loss": 0.0873,
+      "loss": 0.0808,
       "step": 7355
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.351238519343168e-06,
-      "loss": 0.0357,
+      "loss": 0.1226,
       "step": 7356
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.3493830596530296e-06,
-      "loss": 0.1371,
+      "loss": 0.165,
       "step": 7357
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.3475275999628914e-06,
-      "loss": 0.0861,
+      "loss": 0.0644,
       "step": 7358
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.3456721402727525e-06,
-      "loss": 0.1912,
+      "loss": 0.0291,
       "step": 7359
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.343816680582615e-06,
-      "loss": 0.1399,
+      "loss": 0.078,
       "step": 7360
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.341961220892476e-06,
-      "loss": 0.0341,
+      "loss": 0.0727,
       "step": 7361
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.340105761202339e-06,
-      "loss": 0.0342,
+      "loss": 0.0197,
       "step": 7362
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.3382503015122e-06,
-      "loss": 0.1391,
+      "loss": 0.0786,
       "step": 7363
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.336394841822062e-06,
-      "loss": 0.1376,
+      "loss": 0.0398,
       "step": 7364
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.334539382131924e-06,
-      "loss": 0.1371,
+      "loss": 0.0618,
       "step": 7365
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.3326839224417856e-06,
-      "loss": 0.0316,
+      "loss": 0.0862,
       "step": 7366
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.330828462751647e-06,
-      "loss": 0.0859,
+      "loss": 0.0168,
       "step": 7367
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.328973003061509e-06,
-      "loss": 0.0865,
+      "loss": 0.2383,
       "step": 7368
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.32711754337137e-06,
-      "loss": 0.2532,
+      "loss": 0.1025,
       "step": 7369
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.325262083681233e-06,
-      "loss": 0.1431,
+      "loss": 0.0273,
       "step": 7370
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.323406623991094e-06,
-      "loss": 0.2434,
+      "loss": 0.0359,
       "step": 7371
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.321551164300957e-06,
-      "loss": 0.0845,
+      "loss": 0.017,
       "step": 7372
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.319695704610818e-06,
-      "loss": 0.0842,
+      "loss": 0.1077,
       "step": 7373
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.31784024492068e-06,
-      "loss": 0.0861,
+      "loss": 0.0165,
       "step": 7374
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.3159847852305416e-06,
-      "loss": 0.2548,
+      "loss": 0.0134,
       "step": 7375
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.314129325540403e-06,
-      "loss": 0.1389,
+      "loss": 0.0106,
       "step": 7376
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.3122738658502644e-06,
-      "loss": 0.5784,
+      "loss": 0.0611,
       "step": 7377
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.310418406160127e-06,
-      "loss": 0.1957,
+      "loss": 0.1381,
       "step": 7378
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.308562946469988e-06,
-      "loss": 0.1358,
+      "loss": 0.0176,
       "step": 7379
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.306707486779851e-06,
-      "loss": 0.1931,
+      "loss": 0.0084,
       "step": 7380
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.304852027089712e-06,
-      "loss": 0.2537,
+      "loss": 0.0498,
       "step": 7381
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.302996567399574e-06,
-      "loss": 0.1411,
+      "loss": 0.0131,
       "step": 7382
     },
     {
       "epoch": 2.05,
       "learning_rate": 6.301141107709436e-06,
-      "loss": 0.1909,
+      "loss": 0.0155,
       "step": 7383
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.2992856480192975e-06,
-      "loss": 0.195,
+      "loss": 0.1189,
       "step": 7384
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.2974301883291586e-06,
-      "loss": 0.2993,
+      "loss": 0.076,
       "step": 7385
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.295574728639021e-06,
-      "loss": 0.0846,
+      "loss": 0.0492,
       "step": 7386
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.293719268948882e-06,
-      "loss": 0.1383,
+      "loss": 0.0118,
       "step": 7387
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.291863809258745e-06,
-      "loss": 0.302,
+      "loss": 0.0138,
       "step": 7388
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.290008349568606e-06,
-      "loss": 0.1414,
+      "loss": 0.0516,
       "step": 7389
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.288152889878468e-06,
-      "loss": 0.2469,
+      "loss": 0.1025,
       "step": 7390
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.28629743018833e-06,
-      "loss": 0.0847,
+      "loss": 0.0077,
       "step": 7391
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.284441970498192e-06,
-      "loss": 0.1363,
+      "loss": 0.0118,
       "step": 7392
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.282586510808053e-06,
-      "loss": 0.1964,
+      "loss": 0.007,
       "step": 7393
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.280731051117915e-06,
-      "loss": 0.2404,
+      "loss": 0.0135,
       "step": 7394
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.278875591427776e-06,
-      "loss": 0.2423,
+      "loss": 0.0683,
       "step": 7395
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.277020131737639e-06,
-      "loss": 0.1938,
+      "loss": 0.0619,
       "step": 7396
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.2751646720475e-06,
-      "loss": 0.088,
+      "loss": 0.0117,
       "step": 7397
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.273309212357362e-06,
-      "loss": 0.1933,
+      "loss": 0.0068,
       "step": 7398
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.271453752667224e-06,
-      "loss": 0.0874,
+      "loss": 0.0957,
       "step": 7399
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.269598292977086e-06,
-      "loss": 0.0885,
+      "loss": 0.048,
       "step": 7400
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.267742833286947e-06,
-      "loss": 0.1928,
+      "loss": 0.2041,
       "step": 7401
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.2658873735968095e-06,
-      "loss": 0.1389,
+      "loss": 0.0069,
       "step": 7402
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.2640319139066705e-06,
-      "loss": 0.2421,
+      "loss": 0.0088,
       "step": 7403
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.262176454216533e-06,
-      "loss": 0.0876,
+      "loss": 0.0588,
       "step": 7404
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.260320994526394e-06,
-      "loss": 0.0903,
+      "loss": 0.0097,
       "step": 7405
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.258465534836257e-06,
-      "loss": 0.1942,
+      "loss": 0.1397,
       "step": 7406
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.256610075146118e-06,
-      "loss": 0.0901,
+      "loss": 0.0118,
       "step": 7407
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.25475461545598e-06,
-      "loss": 0.1381,
+      "loss": 0.1144,
       "step": 7408
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.252899155765842e-06,
-      "loss": 0.1887,
+      "loss": 0.0649,
       "step": 7409
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.251043696075704e-06,
-      "loss": 0.0899,
+      "loss": 0.0714,
       "step": 7410
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.249188236385565e-06,
-      "loss": 0.0366,
+      "loss": 0.1026,
       "step": 7411
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.247332776695427e-06,
-      "loss": 0.139,
+      "loss": 0.1154,
       "step": 7412
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.245477317005288e-06,
-      "loss": 0.0901,
+      "loss": 0.0101,
       "step": 7413
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.243621857315151e-06,
-      "loss": 0.0901,
+      "loss": 0.0225,
       "step": 7414
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.241766397625012e-06,
-      "loss": 0.0865,
+      "loss": 0.097,
       "step": 7415
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.239910937934874e-06,
-      "loss": 0.1941,
+      "loss": 0.0273,
       "step": 7416
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.238055478244736e-06,
-      "loss": 0.0363,
+      "loss": 0.0513,
       "step": 7417
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.236200018554598e-06,
-      "loss": 0.2412,
+      "loss": 0.0444,
       "step": 7418
     },
     {
       "epoch": 2.06,
       "learning_rate": 6.234344558864459e-06,
-      "loss": 0.0364,
+      "loss": 0.0161,
       "step": 7419
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.2324890991743215e-06,
-      "loss": 0.1374,
+      "loss": 0.0895,
       "step": 7420
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.2306336394841825e-06,
-      "loss": 0.0868,
+      "loss": 0.0155,
       "step": 7421
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.228778179794045e-06,
-      "loss": 0.0881,
+      "loss": 0.0955,
       "step": 7422
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.226922720103906e-06,
-      "loss": 0.0866,
+      "loss": 0.0792,
       "step": 7423
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.225067260413768e-06,
-      "loss": 0.2465,
+      "loss": 0.112,
       "step": 7424
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.22321180072363e-06,
-      "loss": 0.1937,
+      "loss": 0.0397,
       "step": 7425
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.221356341033492e-06,
-      "loss": 0.1394,
+      "loss": 0.059,
       "step": 7426
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.219500881343353e-06,
-      "loss": 0.0332,
+      "loss": 0.064,
       "step": 7427
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.217645421653216e-06,
-      "loss": 0.0873,
+      "loss": 0.1172,
       "step": 7428
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.215789961963077e-06,
-      "loss": 0.2425,
+      "loss": 0.0143,
       "step": 7429
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.213934502272939e-06,
-      "loss": 0.3025,
+      "loss": 0.072,
       "step": 7430
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.2120790425828e-06,
-      "loss": 0.0844,
+      "loss": 0.0198,
       "step": 7431
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.210223582892663e-06,
-      "loss": 0.1366,
+      "loss": 0.0472,
       "step": 7432
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.208368123202524e-06,
-      "loss": 0.1931,
+      "loss": 0.0546,
       "step": 7433
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.206512663512386e-06,
-      "loss": 0.1906,
+      "loss": 0.1158,
       "step": 7434
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.204657203822248e-06,
-      "loss": 0.1385,
+      "loss": 0.1599,
       "step": 7435
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.20280174413211e-06,
-      "loss": 0.0853,
+      "loss": 0.0519,
       "step": 7436
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.200946284441971e-06,
-      "loss": 0.0322,
+      "loss": 0.0107,
       "step": 7437
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.1990908247518335e-06,
-      "loss": 0.2454,
+      "loss": 0.0611,
       "step": 7438
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.1972353650616945e-06,
-      "loss": 0.0853,
+      "loss": 0.0596,
       "step": 7439
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.1953799053715555e-06,
-      "loss": 0.1932,
+      "loss": 0.1596,
       "step": 7440
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.193524445681418e-06,
-      "loss": 0.1395,
+      "loss": 0.043,
       "step": 7441
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.191668985991279e-06,
-      "loss": 0.1857,
+      "loss": 0.0485,
       "step": 7442
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.189813526301142e-06,
-      "loss": 0.1954,
+      "loss": 0.1218,
       "step": 7443
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.187958066611003e-06,
-      "loss": 0.1384,
+      "loss": 0.0149,
       "step": 7444
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.186102606920865e-06,
-      "loss": 0.19,
+      "loss": 0.1267,
       "step": 7445
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.184247147230727e-06,
-      "loss": 0.0335,
+      "loss": 0.0342,
       "step": 7446
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.182391687540589e-06,
-      "loss": 0.0878,
+      "loss": 0.0497,
       "step": 7447
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.18053622785045e-06,
-      "loss": 0.1363,
+      "loss": 0.059,
       "step": 7448
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.178680768160312e-06,
-      "loss": 0.0842,
+      "loss": 0.0652,
       "step": 7449
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.176825308470173e-06,
-      "loss": 0.1891,
+      "loss": 0.0144,
       "step": 7450
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.174969848780036e-06,
-      "loss": 0.2465,
+      "loss": 0.0128,
       "step": 7451
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.173114389089897e-06,
-      "loss": 0.0883,
+      "loss": 0.0215,
       "step": 7452
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.171258929399759e-06,
-      "loss": 0.0865,
+      "loss": 0.021,
       "step": 7453
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.169403469709621e-06,
-      "loss": 0.0317,
+      "loss": 0.0185,
       "step": 7454
     },
     {
       "epoch": 2.07,
       "learning_rate": 6.167548010019483e-06,
-      "loss": 0.1911,
+      "loss": 0.0485,
       "step": 7455
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.165692550329344e-06,
-      "loss": 0.0858,
+      "loss": 0.0099,
       "step": 7456
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.1638370906392065e-06,
-      "loss": 0.1959,
+      "loss": 0.0165,
       "step": 7457
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.1619816309490675e-06,
-      "loss": 0.0829,
+      "loss": 0.018,
       "step": 7458
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.16012617125893e-06,
-      "loss": 0.0329,
+      "loss": 0.0085,
       "step": 7459
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.158270711568791e-06,
-      "loss": 0.1388,
+      "loss": 0.1494,
       "step": 7460
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.156415251878653e-06,
-      "loss": 0.1385,
+      "loss": 0.0435,
       "step": 7461
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.154559792188515e-06,
-      "loss": 0.2476,
+      "loss": 0.1022,
       "step": 7462
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.152704332498377e-06,
-      "loss": 0.1941,
+      "loss": 0.0092,
       "step": 7463
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.150848872808238e-06,
-      "loss": 0.0867,
+      "loss": 0.1341,
       "step": 7464
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.148993413118101e-06,
-      "loss": 0.0853,
+      "loss": 0.203,
       "step": 7465
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.147137953427962e-06,
-      "loss": 0.1898,
+      "loss": 0.2062,
       "step": 7466
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.145282493737824e-06,
-      "loss": 0.0855,
+      "loss": 0.0678,
       "step": 7467
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.143427034047685e-06,
-      "loss": 0.0839,
+      "loss": 0.0555,
       "step": 7468
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.141571574357548e-06,
-      "loss": 0.1919,
+      "loss": 0.1189,
       "step": 7469
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.139716114667409e-06,
-      "loss": 0.1972,
+      "loss": 0.1256,
       "step": 7470
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.137860654977271e-06,
-      "loss": 0.191,
+      "loss": 0.0606,
       "step": 7471
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.136005195287133e-06,
-      "loss": 0.031,
+      "loss": 0.1702,
       "step": 7472
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.134149735596995e-06,
-      "loss": 0.0316,
+      "loss": 0.1014,
       "step": 7473
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.132294275906856e-06,
-      "loss": 0.1413,
+      "loss": 0.0516,
       "step": 7474
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.1304388162167184e-06,
-      "loss": 0.2964,
+      "loss": 0.0515,
       "step": 7475
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.1285833565265795e-06,
-      "loss": 0.2442,
+      "loss": 0.0816,
       "step": 7476
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.126727896836442e-06,
-      "loss": 0.139,
+      "loss": 0.0835,
       "step": 7477
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.124872437146303e-06,
-      "loss": 0.1398,
+      "loss": 0.0236,
       "step": 7478
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.123016977456165e-06,
-      "loss": 0.1371,
+      "loss": 0.1193,
       "step": 7479
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.121161517766027e-06,
-      "loss": 0.248,
+      "loss": 0.0905,
       "step": 7480
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.119306058075889e-06,
-      "loss": 0.1398,
+      "loss": 0.0207,
       "step": 7481
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.11745059838575e-06,
-      "loss": 0.1906,
+      "loss": 0.0979,
       "step": 7482
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.1155951386956126e-06,
-      "loss": 0.1334,
+      "loss": 0.208,
       "step": 7483
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.113739679005474e-06,
-      "loss": 0.0845,
+      "loss": 0.0252,
       "step": 7484
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.111884219315336e-06,
-      "loss": 0.1959,
+      "loss": 0.0581,
       "step": 7485
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.110028759625197e-06,
-      "loss": 0.0331,
+      "loss": 0.1254,
       "step": 7486
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.108173299935059e-06,
-      "loss": 0.0886,
+      "loss": 0.1092,
       "step": 7487
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.106317840244921e-06,
-      "loss": 0.2417,
+      "loss": 0.0622,
       "step": 7488
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.104462380554783e-06,
-      "loss": 0.2927,
+      "loss": 0.0654,
       "step": 7489
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.102606920864644e-06,
-      "loss": 0.1935,
+      "loss": 0.1714,
       "step": 7490
     },
     {
       "epoch": 2.08,
       "learning_rate": 6.100751461174507e-06,
-      "loss": 0.0847,
+      "loss": 0.129,
       "step": 7491
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.098896001484368e-06,
-      "loss": 0.0879,
+      "loss": 0.1237,
       "step": 7492
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.09704054179423e-06,
-      "loss": 0.1391,
+      "loss": 0.0225,
       "step": 7493
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.0951850821040914e-06,
-      "loss": 0.3002,
+      "loss": 0.0746,
       "step": 7494
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.093329622413954e-06,
-      "loss": 0.0882,
+      "loss": 0.0899,
       "step": 7495
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.091474162723815e-06,
-      "loss": 0.1918,
+      "loss": 0.0373,
       "step": 7496
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.089618703033677e-06,
-      "loss": 0.1392,
+      "loss": 0.0989,
       "step": 7497
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.087763243343539e-06,
-      "loss": 0.0853,
+      "loss": 0.1186,
       "step": 7498
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.085907783653401e-06,
-      "loss": 0.0343,
+      "loss": 0.0577,
       "step": 7499
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.084052323963262e-06,
-      "loss": 0.1352,
+      "loss": 0.0949,
       "step": 7500
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.0821968642731245e-06,
-      "loss": 0.2946,
+      "loss": 0.0549,
       "step": 7501
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.0803414045829856e-06,
-      "loss": 0.1418,
+      "loss": 0.0211,
       "step": 7502
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.078485944892848e-06,
-      "loss": 0.1889,
+      "loss": 0.1039,
       "step": 7503
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.076630485202709e-06,
-      "loss": 0.1386,
+      "loss": 0.1255,
       "step": 7504
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.074775025512571e-06,
-      "loss": 0.2364,
+      "loss": 0.1229,
       "step": 7505
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.072919565822433e-06,
-      "loss": 0.0347,
+      "loss": 0.1135,
       "step": 7506
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.071064106132295e-06,
-      "loss": 0.1428,
+      "loss": 0.0542,
       "step": 7507
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.069208646442156e-06,
-      "loss": 0.0874,
+      "loss": 0.0231,
       "step": 7508
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.067353186752019e-06,
-      "loss": 0.1369,
+      "loss": 0.0729,
       "step": 7509
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.06549772706188e-06,
-      "loss": 0.1411,
+      "loss": 0.0851,
       "step": 7510
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.063642267371742e-06,
-      "loss": 0.1925,
+      "loss": 0.1615,
       "step": 7511
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.061786807681603e-06,
-      "loss": 0.1967,
+      "loss": 0.0217,
       "step": 7512
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.059931347991465e-06,
-      "loss": 0.1389,
+      "loss": 0.084,
       "step": 7513
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.058075888301327e-06,
-      "loss": 0.0346,
+      "loss": 0.0658,
       "step": 7514
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.056220428611189e-06,
-      "loss": 0.0851,
+      "loss": 0.013,
       "step": 7515
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.05436496892105e-06,
-      "loss": 0.0859,
+      "loss": 0.071,
       "step": 7516
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.052509509230913e-06,
-      "loss": 0.2499,
+      "loss": 0.0747,
       "step": 7517
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.050654049540774e-06,
-      "loss": 0.2417,
+      "loss": 0.1424,
       "step": 7518
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.0487985898506365e-06,
-      "loss": 0.0869,
+      "loss": 0.0247,
       "step": 7519
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.0469431301604975e-06,
-      "loss": 0.2428,
+      "loss": 0.0251,
       "step": 7520
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.045087670470359e-06,
-      "loss": 0.2497,
+      "loss": 0.1345,
       "step": 7521
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.043232210780221e-06,
-      "loss": 0.0884,
+      "loss": 0.0241,
       "step": 7522
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.041376751090083e-06,
-      "loss": 0.1891,
+      "loss": 0.0563,
       "step": 7523
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.039521291399944e-06,
-      "loss": 0.034,
+      "loss": 0.0472,
       "step": 7524
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.037665831709807e-06,
-      "loss": 0.1419,
+      "loss": 0.1332,
       "step": 7525
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.035810372019668e-06,
-      "loss": 0.1876,
+      "loss": 0.0588,
       "step": 7526
     },
     {
       "epoch": 2.09,
       "learning_rate": 6.033954912329531e-06,
-      "loss": 0.0851,
+      "loss": 0.0509,
       "step": 7527
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.032099452639392e-06,
-      "loss": 0.2474,
+      "loss": 0.0476,
       "step": 7528
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.030243992949254e-06,
-      "loss": 0.2366,
+      "loss": 0.0552,
       "step": 7529
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.028388533259115e-06,
-      "loss": 0.2432,
+      "loss": 0.0475,
       "step": 7530
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.026533073568977e-06,
-      "loss": 0.1917,
+      "loss": 0.1975,
       "step": 7531
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.024677613878839e-06,
-      "loss": 0.2472,
+      "loss": 0.0185,
       "step": 7532
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.022822154188701e-06,
-      "loss": 0.0863,
+      "loss": 0.0349,
       "step": 7533
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.020966694498562e-06,
-      "loss": 0.1368,
+      "loss": 0.0729,
       "step": 7534
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.019111234808425e-06,
-      "loss": 0.0868,
+      "loss": 0.0733,
       "step": 7535
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.017255775118286e-06,
-      "loss": 0.1399,
+      "loss": 0.0115,
       "step": 7536
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.0154003154281485e-06,
-      "loss": 0.0861,
+      "loss": 0.015,
       "step": 7537
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.0135448557380095e-06,
-      "loss": 0.2407,
+      "loss": 0.077,
       "step": 7538
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.011689396047871e-06,
-      "loss": 0.1919,
+      "loss": 0.1113,
       "step": 7539
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.009833936357733e-06,
-      "loss": 0.088,
+      "loss": 0.0169,
       "step": 7540
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.007978476667595e-06,
-      "loss": 0.1392,
+      "loss": 0.0089,
       "step": 7541
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.006123016977456e-06,
-      "loss": 0.1937,
+      "loss": 0.2588,
       "step": 7542
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.004267557287319e-06,
-      "loss": 0.1904,
+      "loss": 0.1946,
       "step": 7543
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.00241209759718e-06,
-      "loss": 0.1896,
+      "loss": 0.0616,
       "step": 7544
     },
     {
       "epoch": 2.1,
       "learning_rate": 6.000556637907043e-06,
-      "loss": 0.1878,
+      "loss": 0.0483,
       "step": 7545
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.998701178216904e-06,
-      "loss": 0.1907,
+      "loss": 0.0188,
       "step": 7546
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.9968457185267655e-06,
-      "loss": 0.1387,
+      "loss": 0.0088,
       "step": 7547
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.994990258836627e-06,
-      "loss": 0.1387,
+      "loss": 0.1047,
       "step": 7548
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.993134799146489e-06,
-      "loss": 0.2348,
+      "loss": 0.0102,
       "step": 7549
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.99127933945635e-06,
-      "loss": 0.1408,
+      "loss": 0.0926,
       "step": 7550
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.989423879766213e-06,
-      "loss": 0.1895,
+      "loss": 0.1161,
       "step": 7551
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.987568420076074e-06,
-      "loss": 0.0891,
+      "loss": 0.1685,
       "step": 7552
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.985712960385937e-06,
-      "loss": 0.2436,
+      "loss": 0.0191,
       "step": 7553
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.983857500695798e-06,
-      "loss": 0.1404,
+      "loss": 0.0169,
       "step": 7554
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.9820020410056605e-06,
-      "loss": 0.2429,
+      "loss": 0.0513,
       "step": 7555
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.9801465813155215e-06,
-      "loss": 0.137,
+      "loss": 0.1059,
       "step": 7556
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.978291121625383e-06,
-      "loss": 0.1905,
+      "loss": 0.1309,
       "step": 7557
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.976435661935245e-06,
-      "loss": 0.1385,
+      "loss": 0.058,
       "step": 7558
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.974580202245107e-06,
-      "loss": 0.1441,
+      "loss": 0.1001,
       "step": 7559
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.972724742554968e-06,
-      "loss": 0.1917,
+      "loss": 0.0194,
       "step": 7560
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.970869282864831e-06,
-      "loss": 0.0895,
+      "loss": 0.0937,
       "step": 7561
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.969013823174692e-06,
-      "loss": 0.1373,
+      "loss": 0.0102,
       "step": 7562
     },
     {
       "epoch": 2.1,
       "learning_rate": 5.967158363484555e-06,
-      "loss": 0.1894,
+      "loss": 0.0264,
       "step": 7563
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.965302903794416e-06,
-      "loss": 0.242,
+      "loss": 0.0554,
       "step": 7564
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.9634474441042775e-06,
-      "loss": 0.1901,
+      "loss": 0.1338,
       "step": 7565
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.961591984414139e-06,
-      "loss": 0.2384,
+      "loss": 0.0491,
       "step": 7566
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.959736524724001e-06,
-      "loss": 0.0408,
+      "loss": 0.1536,
       "step": 7567
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.957881065033862e-06,
-      "loss": 0.0901,
+      "loss": 0.1554,
       "step": 7568
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.956025605343725e-06,
-      "loss": 0.1381,
+      "loss": 0.1638,
       "step": 7569
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.954170145653586e-06,
-      "loss": 0.1337,
+      "loss": 0.061,
       "step": 7570
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.952314685963449e-06,
-      "loss": 0.0387,
+      "loss": 0.1386,
       "step": 7571
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.95045922627331e-06,
-      "loss": 0.0908,
+      "loss": 0.0627,
       "step": 7572
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.948603766583172e-06,
-      "loss": 0.0895,
+      "loss": 0.0172,
       "step": 7573
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.9467483068930335e-06,
-      "loss": 0.0892,
+      "loss": 0.0667,
       "step": 7574
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.944892847202895e-06,
-      "loss": 0.0897,
+      "loss": 0.0231,
       "step": 7575
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.943037387512756e-06,
-      "loss": 0.2381,
+      "loss": 0.13,
       "step": 7576
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.941181927822619e-06,
-      "loss": 0.1424,
+      "loss": 0.0234,
       "step": 7577
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.93932646813248e-06,
-      "loss": 0.0871,
+      "loss": 0.0553,
       "step": 7578
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.937471008442343e-06,
-      "loss": 0.1904,
+      "loss": 0.052,
       "step": 7579
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.935615548752204e-06,
-      "loss": 0.2413,
+      "loss": 0.0488,
       "step": 7580
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.933760089062066e-06,
-      "loss": 0.2941,
+      "loss": 0.0206,
       "step": 7581
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.931904629371928e-06,
-      "loss": 0.1386,
+      "loss": 0.0958,
       "step": 7582
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.9300491696817895e-06,
-      "loss": 0.1434,
+      "loss": 0.0188,
       "step": 7583
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.9281937099916505e-06,
-      "loss": 0.0862,
+      "loss": 0.1122,
       "step": 7584
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.926338250301513e-06,
-      "loss": 0.1404,
+      "loss": 0.0211,
       "step": 7585
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.924482790611374e-06,
-      "loss": 0.0358,
+      "loss": 0.0837,
       "step": 7586
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.922627330921237e-06,
-      "loss": 0.1396,
+      "loss": 0.1035,
       "step": 7587
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.920771871231098e-06,
-      "loss": 0.0865,
+      "loss": 0.2022,
       "step": 7588
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.918916411540961e-06,
-      "loss": 0.0883,
+      "loss": 0.0374,
       "step": 7589
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.917060951850822e-06,
-      "loss": 0.2385,
+      "loss": 0.0126,
       "step": 7590
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.915205492160683e-06,
-      "loss": 0.0893,
+      "loss": 0.0109,
       "step": 7591
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.9133500324705454e-06,
-      "loss": 0.1387,
+      "loss": 0.1413,
       "step": 7592
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.9114945727804065e-06,
-      "loss": 0.1418,
+      "loss": 0.0631,
       "step": 7593
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.909639113090268e-06,
-      "loss": 0.0878,
+      "loss": 0.0848,
       "step": 7594
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.90778365340013e-06,
-      "loss": 0.089,
+      "loss": 0.1477,
       "step": 7595
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.905928193709992e-06,
-      "loss": 0.035,
+      "loss": 0.0507,
       "step": 7596
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.904072734019853e-06,
-      "loss": 0.2981,
+      "loss": 0.086,
       "step": 7597
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.902217274329716e-06,
-      "loss": 0.1399,
+      "loss": 0.0168,
       "step": 7598
     },
     {
       "epoch": 2.11,
       "learning_rate": 5.900361814639577e-06,
-      "loss": 0.2943,
+      "loss": 0.2083,
       "step": 7599
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.8985063549494396e-06,
-      "loss": 0.0864,
+      "loss": 0.094,
       "step": 7600
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.896650895259301e-06,
-      "loss": 0.0335,
+      "loss": 0.1193,
       "step": 7601
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.8947954355691625e-06,
-      "loss": 0.1892,
+      "loss": 0.0815,
       "step": 7602
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.892939975879024e-06,
-      "loss": 0.0904,
+      "loss": 0.1415,
       "step": 7603
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.891084516188886e-06,
-      "loss": 0.1962,
+      "loss": 0.0162,
       "step": 7604
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.889229056498747e-06,
-      "loss": 0.1407,
+      "loss": 0.1805,
       "step": 7605
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.88737359680861e-06,
-      "loss": 0.3017,
+      "loss": 0.0799,
       "step": 7606
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.885518137118471e-06,
-      "loss": 0.1357,
+      "loss": 0.166,
       "step": 7607
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.883662677428334e-06,
-      "loss": 0.0855,
+      "loss": 0.019,
       "step": 7608
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.881807217738195e-06,
-      "loss": 0.0322,
+      "loss": 0.0207,
       "step": 7609
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.8799517580480566e-06,
-      "loss": 0.1394,
+      "loss": 0.11,
       "step": 7610
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.8780962983579184e-06,
-      "loss": 0.1421,
+      "loss": 0.0566,
       "step": 7611
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.87624083866778e-06,
-      "loss": 0.1414,
+      "loss": 0.0252,
       "step": 7612
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.874385378977641e-06,
-      "loss": 0.0842,
+      "loss": 0.0493,
       "step": 7613
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.872529919287504e-06,
-      "loss": 0.1393,
+      "loss": 0.0783,
       "step": 7614
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.870674459597365e-06,
-      "loss": 0.2477,
+      "loss": 0.0704,
       "step": 7615
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.868818999907228e-06,
-      "loss": 0.1921,
+      "loss": 0.0462,
       "step": 7616
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.866963540217089e-06,
-      "loss": 0.2462,
+      "loss": 0.023,
       "step": 7617
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.8651080805269515e-06,
-      "loss": 0.1386,
+      "loss": 0.067,
       "step": 7618
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.8632526208368126e-06,
-      "loss": 0.1919,
+      "loss": 0.0294,
       "step": 7619
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.8613971611466744e-06,
-      "loss": 0.1374,
+      "loss": 0.0683,
       "step": 7620
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.859541701456536e-06,
-      "loss": 0.0842,
+      "loss": 0.0486,
       "step": 7621
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.857686241766398e-06,
-      "loss": 0.0335,
+      "loss": 0.0913,
       "step": 7622
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.855830782076259e-06,
-      "loss": 0.1383,
+      "loss": 0.0882,
       "step": 7623
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.853975322386122e-06,
-      "loss": 0.0847,
+      "loss": 0.0471,
       "step": 7624
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.852119862695983e-06,
-      "loss": 0.0325,
+      "loss": 0.0147,
       "step": 7625
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.850264403005846e-06,
-      "loss": 0.1406,
+      "loss": 0.0501,
       "step": 7626
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.848408943315707e-06,
-      "loss": 0.0886,
+      "loss": 0.0891,
       "step": 7627
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.8465534836255685e-06,
-      "loss": 0.2455,
+      "loss": 0.0673,
       "step": 7628
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.84469802393543e-06,
-      "loss": 0.1399,
+      "loss": 0.0235,
       "step": 7629
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.842842564245292e-06,
-      "loss": 0.0318,
+      "loss": 0.1679,
       "step": 7630
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.840987104555153e-06,
-      "loss": 0.0857,
+      "loss": 0.0832,
       "step": 7631
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.839131644865016e-06,
-      "loss": 0.1404,
+      "loss": 0.12,
       "step": 7632
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.837276185174877e-06,
-      "loss": 0.1939,
+      "loss": 0.0447,
       "step": 7633
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.83542072548474e-06,
-      "loss": 0.3525,
+      "loss": 0.0486,
       "step": 7634
     },
     {
       "epoch": 2.12,
       "learning_rate": 5.833565265794601e-06,
-      "loss": 0.1366,
+      "loss": 0.1563,
       "step": 7635
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.831709806104463e-06,
-      "loss": 0.0318,
+      "loss": 0.1305,
       "step": 7636
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.8298543464143245e-06,
-      "loss": 0.0882,
+      "loss": 0.1004,
       "step": 7637
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.827998886724186e-06,
-      "loss": 0.138,
+      "loss": 0.0533,
       "step": 7638
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.8261434270340474e-06,
-      "loss": 0.1384,
+      "loss": 0.0935,
       "step": 7639
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.82428796734391e-06,
-      "loss": 0.0846,
+      "loss": 0.0647,
       "step": 7640
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.822432507653771e-06,
-      "loss": 0.0317,
+      "loss": 0.1755,
       "step": 7641
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.820577047963634e-06,
-      "loss": 0.2431,
+      "loss": 0.0381,
       "step": 7642
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.818721588273495e-06,
-      "loss": 0.1427,
+      "loss": 0.0613,
       "step": 7643
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.816866128583357e-06,
-      "loss": 0.2494,
+      "loss": 0.1657,
       "step": 7644
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.815010668893219e-06,
-      "loss": 0.0833,
+      "loss": 0.0671,
       "step": 7645
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.8131552092030805e-06,
-      "loss": 0.1375,
+      "loss": 0.0718,
       "step": 7646
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.8112997495129415e-06,
-      "loss": 0.0309,
+      "loss": 0.0797,
       "step": 7647
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.809444289822804e-06,
-      "loss": 0.1376,
+      "loss": 0.0294,
       "step": 7648
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.807588830132665e-06,
-      "loss": 0.0307,
+      "loss": 0.0187,
       "step": 7649
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.805733370442528e-06,
-      "loss": 0.242,
+      "loss": 0.0292,
       "step": 7650
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.803877910752389e-06,
-      "loss": 0.1402,
+      "loss": 0.0487,
       "step": 7651
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.802022451062252e-06,
-      "loss": 0.1414,
+      "loss": 0.0428,
       "step": 7652
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.800166991372113e-06,
-      "loss": 0.1997,
+      "loss": 0.0569,
       "step": 7653
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.798311531681975e-06,
-      "loss": 0.1929,
+      "loss": 0.2109,
       "step": 7654
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.7964560719918365e-06,
-      "loss": 0.14,
+      "loss": 0.087,
       "step": 7655
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.794600612301698e-06,
-      "loss": 0.0311,
+      "loss": 0.0218,
       "step": 7656
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.792745152611559e-06,
-      "loss": 0.0875,
+      "loss": 0.0607,
       "step": 7657
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.790889692921422e-06,
-      "loss": 0.2467,
+      "loss": 0.1016,
       "step": 7658
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.789034233231283e-06,
-      "loss": 0.1381,
+      "loss": 0.1111,
       "step": 7659
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.787178773541146e-06,
-      "loss": 0.2459,
+      "loss": 0.0264,
       "step": 7660
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.785323313851007e-06,
-      "loss": 0.084,
+      "loss": 0.0329,
       "step": 7661
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.783467854160869e-06,
-      "loss": 0.0855,
+      "loss": 0.0178,
       "step": 7662
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.781612394470731e-06,
-      "loss": 0.1394,
+      "loss": 0.0596,
       "step": 7663
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.7797569347805925e-06,
-      "loss": 0.2468,
+      "loss": 0.0398,
       "step": 7664
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.7779014750904535e-06,
-      "loss": 0.0858,
+      "loss": 0.1004,
       "step": 7665
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.776046015400316e-06,
-      "loss": 0.0852,
+      "loss": 0.1016,
       "step": 7666
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.774190555710177e-06,
-      "loss": 0.1902,
+      "loss": 0.0093,
       "step": 7667
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.77233509602004e-06,
-      "loss": 0.14,
+      "loss": 0.1186,
       "step": 7668
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.770479636329901e-06,
-      "loss": 0.1891,
+      "loss": 0.1313,
       "step": 7669
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.768624176639763e-06,
-      "loss": 0.1417,
+      "loss": 0.0976,
       "step": 7670
     },
     {
       "epoch": 2.13,
       "learning_rate": 5.766768716949625e-06,
-      "loss": 0.1379,
+      "loss": 0.1905,
       "step": 7671
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.764913257259487e-06,
-      "loss": 0.2402,
+      "loss": 0.0514,
       "step": 7672
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.763057797569348e-06,
-      "loss": 0.1963,
+      "loss": 0.0194,
       "step": 7673
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.76120233787921e-06,
-      "loss": 0.0842,
+      "loss": 0.0239,
       "step": 7674
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.759346878189071e-06,
-      "loss": 0.1403,
+      "loss": 0.0471,
       "step": 7675
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.757491418498934e-06,
-      "loss": 0.1419,
+      "loss": 0.05,
       "step": 7676
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.755635958808795e-06,
-      "loss": 0.1887,
+      "loss": 0.1443,
       "step": 7677
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.753780499118658e-06,
-      "loss": 0.1984,
+      "loss": 0.0266,
       "step": 7678
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.751925039428519e-06,
-      "loss": 0.1924,
+      "loss": 0.0491,
       "step": 7679
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.750069579738381e-06,
-      "loss": 0.1425,
+      "loss": 0.1076,
       "step": 7680
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.748214120048243e-06,
-      "loss": 0.1399,
+      "loss": 0.0925,
       "step": 7681
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.7463586603581045e-06,
-      "loss": 0.0856,
+      "loss": 0.0272,
       "step": 7682
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.7445032006679655e-06,
-      "loss": 0.1397,
+      "loss": 0.0505,
       "step": 7683
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.742647740977828e-06,
-      "loss": 0.0862,
+      "loss": 0.0353,
       "step": 7684
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.740792281287689e-06,
-      "loss": 0.1894,
+      "loss": 0.1228,
       "step": 7685
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.738936821597552e-06,
-      "loss": 0.1392,
+      "loss": 0.0186,
       "step": 7686
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.737081361907413e-06,
-      "loss": 0.1399,
+      "loss": 0.0345,
       "step": 7687
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.735225902217275e-06,
-      "loss": 0.033,
+      "loss": 0.0172,
       "step": 7688
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.733370442527137e-06,
-      "loss": 0.0343,
+      "loss": 0.1092,
       "step": 7689
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.731514982836999e-06,
-      "loss": 0.034,
+      "loss": 0.0216,
       "step": 7690
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.72965952314686e-06,
-      "loss": 0.086,
+      "loss": 0.016,
       "step": 7691
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.727804063456722e-06,
-      "loss": 0.0861,
+      "loss": 0.0835,
       "step": 7692
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.725948603766583e-06,
-      "loss": 0.1936,
+      "loss": 0.1531,
       "step": 7693
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.724093144076446e-06,
-      "loss": 0.0321,
+      "loss": 0.1018,
       "step": 7694
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.722237684386307e-06,
-      "loss": 0.0864,
+      "loss": 0.0122,
       "step": 7695
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.720382224696169e-06,
-      "loss": 0.0865,
+      "loss": 0.0169,
       "step": 7696
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.718526765006031e-06,
-      "loss": 0.245,
+      "loss": 0.1255,
       "step": 7697
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.716671305315893e-06,
-      "loss": 0.0866,
+      "loss": 0.0128,
       "step": 7698
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.714815845625754e-06,
-      "loss": 0.2427,
+      "loss": 0.1681,
       "step": 7699
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.7129603859356165e-06,
-      "loss": 0.0846,
+      "loss": 0.1626,
       "step": 7700
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.7111049262454775e-06,
-      "loss": 0.1401,
+      "loss": 0.1295,
       "step": 7701
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.70924946655534e-06,
-      "loss": 0.1398,
+      "loss": 0.0421,
       "step": 7702
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.707394006865201e-06,
-      "loss": 0.1961,
+      "loss": 0.0089,
       "step": 7703
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.705538547175063e-06,
-      "loss": 0.195,
+      "loss": 0.1728,
       "step": 7704
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.703683087484925e-06,
-      "loss": 0.0833,
+      "loss": 0.0501,
       "step": 7705
     },
     {
       "epoch": 2.14,
       "learning_rate": 5.701827627794787e-06,
-      "loss": 0.1382,
+      "loss": 0.0681,
       "step": 7706
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.699972168104648e-06,
-      "loss": 0.1395,
+      "loss": 0.165,
       "step": 7707
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.6981167084145106e-06,
-      "loss": 0.1387,
+      "loss": 0.0889,
       "step": 7708
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.696261248724372e-06,
-      "loss": 0.0867,
+      "loss": 0.0116,
       "step": 7709
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.694405789034234e-06,
-      "loss": 0.1412,
+      "loss": 0.15,
       "step": 7710
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.692550329344095e-06,
-      "loss": 0.0305,
+      "loss": 0.0647,
       "step": 7711
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.690694869653958e-06,
-      "loss": 0.14,
+      "loss": 0.1062,
       "step": 7712
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.688839409963819e-06,
-      "loss": 0.1386,
+      "loss": 0.022,
       "step": 7713
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.686983950273681e-06,
-      "loss": 0.0302,
+      "loss": 0.1263,
       "step": 7714
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.685128490583543e-06,
-      "loss": 0.1926,
+      "loss": 0.0612,
       "step": 7715
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.683273030893405e-06,
-      "loss": 0.0863,
+      "loss": 0.0763,
       "step": 7716
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.681417571203266e-06,
-      "loss": 0.1408,
+      "loss": 0.0521,
       "step": 7717
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.6795621115131284e-06,
-      "loss": 0.1416,
+      "loss": 0.0713,
       "step": 7718
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.6777066518229895e-06,
-      "loss": 0.0297,
+      "loss": 0.067,
       "step": 7719
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.675851192132852e-06,
-      "loss": 0.0293,
+      "loss": 0.0229,
       "step": 7720
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.673995732442713e-06,
-      "loss": 0.1407,
+      "loss": 0.1268,
       "step": 7721
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.672140272752575e-06,
-      "loss": 0.2483,
+      "loss": 0.13,
       "step": 7722
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.670284813062437e-06,
-      "loss": 0.3072,
+      "loss": 0.0349,
       "step": 7723
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.668429353372299e-06,
-      "loss": 0.0849,
+      "loss": 0.0975,
       "step": 7724
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.66657389368216e-06,
-      "loss": 0.1386,
+      "loss": 0.1491,
       "step": 7725
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.6647184339920225e-06,
-      "loss": 0.0811,
+      "loss": 0.1306,
       "step": 7726
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.6628629743018836e-06,
-      "loss": 0.0825,
+      "loss": 0.1485,
       "step": 7727
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.661007514611746e-06,
-      "loss": 0.1947,
+      "loss": 0.0499,
       "step": 7728
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.659152054921607e-06,
-      "loss": 0.1376,
+      "loss": 0.0249,
       "step": 7729
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.657296595231469e-06,
-      "loss": 0.1938,
+      "loss": 0.1742,
       "step": 7730
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.655441135541331e-06,
-      "loss": 0.0837,
+      "loss": 0.0285,
       "step": 7731
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.653585675851193e-06,
-      "loss": 0.0823,
+      "loss": 0.0506,
       "step": 7732
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.651730216161054e-06,
-      "loss": 0.1431,
+      "loss": 0.0242,
       "step": 7733
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.649874756470917e-06,
-      "loss": 0.1931,
+      "loss": 0.0822,
       "step": 7734
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.648019296780778e-06,
-      "loss": 0.0289,
+      "loss": 0.0259,
       "step": 7735
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.64616383709064e-06,
-      "loss": 0.0823,
+      "loss": 0.0667,
       "step": 7736
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.6443083774005014e-06,
-      "loss": 0.1927,
+      "loss": 0.0169,
       "step": 7737
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.642452917710364e-06,
-      "loss": 0.2497,
+      "loss": 0.0788,
       "step": 7738
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.640597458020225e-06,
-      "loss": 0.2502,
+      "loss": 0.0145,
       "step": 7739
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.638741998330087e-06,
-      "loss": 0.0827,
+      "loss": 0.0222,
       "step": 7740
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.636886538639949e-06,
-      "loss": 0.0849,
+      "loss": 0.0505,
       "step": 7741
     },
     {
       "epoch": 2.15,
       "learning_rate": 5.635031078949811e-06,
-      "loss": 0.084,
+      "loss": 0.0648,
       "step": 7742
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.633175619259672e-06,
-      "loss": 0.029,
+      "loss": 0.1813,
       "step": 7743
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.631320159569534e-06,
-      "loss": 0.1372,
+      "loss": 0.0471,
       "step": 7744
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.6294646998793955e-06,
-      "loss": 0.1947,
+      "loss": 0.0131,
       "step": 7745
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.6276092401892566e-06,
-      "loss": 0.0845,
+      "loss": 0.0701,
       "step": 7746
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.625753780499119e-06,
-      "loss": 0.0832,
+      "loss": 0.0122,
       "step": 7747
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.62389832080898e-06,
-      "loss": 0.1392,
+      "loss": 0.1261,
       "step": 7748
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.622042861118843e-06,
-      "loss": 0.1375,
+      "loss": 0.0929,
       "step": 7749
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.620187401428704e-06,
-      "loss": 0.1872,
+      "loss": 0.0583,
       "step": 7750
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.618331941738566e-06,
-      "loss": 0.0846,
+      "loss": 0.0341,
       "step": 7751
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.616476482048428e-06,
-      "loss": 0.0869,
+      "loss": 0.0633,
       "step": 7752
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.61462102235829e-06,
-      "loss": 0.1938,
+      "loss": 0.0469,
       "step": 7753
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.612765562668151e-06,
-      "loss": 0.0847,
+      "loss": 0.0169,
       "step": 7754
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.610910102978013e-06,
-      "loss": 0.1966,
+      "loss": 0.0659,
       "step": 7755
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.6090546432878744e-06,
-      "loss": 0.1956,
+      "loss": 0.0988,
       "step": 7756
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.607199183597737e-06,
-      "loss": 0.0856,
+      "loss": 0.018,
       "step": 7757
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.605343723907598e-06,
-      "loss": 0.083,
+      "loss": 0.0087,
       "step": 7758
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.60348826421746e-06,
-      "loss": 0.0831,
+      "loss": 0.034,
       "step": 7759
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.601632804527322e-06,
-      "loss": 0.0829,
+      "loss": 0.0671,
       "step": 7760
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.599777344837184e-06,
-      "loss": 0.0851,
+      "loss": 0.0515,
       "step": 7761
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.597921885147045e-06,
-      "loss": 0.1396,
+      "loss": 0.0111,
       "step": 7762
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.5960664254569075e-06,
-      "loss": 0.4116,
+      "loss": 0.0835,
       "step": 7763
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.5942109657667685e-06,
-      "loss": 0.0857,
+      "loss": 0.1849,
       "step": 7764
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.592355506076631e-06,
-      "loss": 0.2505,
+      "loss": 0.2435,
       "step": 7765
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.590500046386492e-06,
-      "loss": 0.0303,
+      "loss": 0.0049,
       "step": 7766
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.588644586696354e-06,
-      "loss": 0.2441,
+      "loss": 0.0809,
       "step": 7767
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.586789127006216e-06,
-      "loss": 0.1976,
+      "loss": 0.0626,
       "step": 7768
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.584933667316078e-06,
-      "loss": 0.2495,
+      "loss": 0.1003,
       "step": 7769
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.583078207625939e-06,
-      "loss": 0.1921,
+      "loss": 0.1134,
       "step": 7770
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.581222747935802e-06,
-      "loss": 0.0313,
+      "loss": 0.0515,
       "step": 7771
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.579367288245663e-06,
-      "loss": 0.1388,
+      "loss": 0.0108,
       "step": 7772
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.577511828555525e-06,
-      "loss": 0.1948,
+      "loss": 0.0204,
       "step": 7773
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.575656368865386e-06,
-      "loss": 0.1924,
+      "loss": 0.0574,
       "step": 7774
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.573800909175249e-06,
-      "loss": 0.4076,
+      "loss": 0.0677,
       "step": 7775
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.57194544948511e-06,
-      "loss": 0.0856,
+      "loss": 0.0181,
       "step": 7776
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.570089989794972e-06,
-      "loss": 0.2463,
+      "loss": 0.1004,
       "step": 7777
     },
     {
       "epoch": 2.16,
       "learning_rate": 5.568234530104834e-06,
-      "loss": 0.0834,
+      "loss": 0.0791,
       "step": 7778
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.566379070414696e-06,
-      "loss": 0.0884,
+      "loss": 0.0183,
       "step": 7779
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.564523610724557e-06,
-      "loss": 0.1414,
+      "loss": 0.0655,
       "step": 7780
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.5626681510344195e-06,
-      "loss": 0.0318,
+      "loss": 0.0158,
       "step": 7781
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.5608126913442805e-06,
-      "loss": 0.0843,
+      "loss": 0.0134,
       "step": 7782
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.558957231654143e-06,
-      "loss": 0.0814,
+      "loss": 0.0175,
       "step": 7783
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.557101771964004e-06,
-      "loss": 0.1921,
+      "loss": 0.0249,
       "step": 7784
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.555246312273866e-06,
-      "loss": 0.2496,
+      "loss": 0.0156,
       "step": 7785
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.553390852583728e-06,
-      "loss": 0.0864,
+      "loss": 0.0173,
       "step": 7786
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.55153539289359e-06,
-      "loss": 0.0321,
+      "loss": 0.1873,
       "step": 7787
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.549679933203451e-06,
-      "loss": 0.0854,
+      "loss": 0.1751,
       "step": 7788
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.547824473513314e-06,
-      "loss": 0.1865,
+      "loss": 0.0536,
       "step": 7789
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.545969013823175e-06,
-      "loss": 0.1392,
+      "loss": 0.01,
       "step": 7790
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.544113554133037e-06,
-      "loss": 0.091,
+      "loss": 0.1368,
       "step": 7791
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.542258094442898e-06,
-      "loss": 0.1959,
+      "loss": 0.0844,
       "step": 7792
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.54040263475276e-06,
-      "loss": 0.1422,
+      "loss": 0.0719,
       "step": 7793
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.538547175062622e-06,
-      "loss": 0.1878,
+      "loss": 0.0539,
       "step": 7794
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.536691715372484e-06,
-      "loss": 0.1896,
+      "loss": 0.0586,
       "step": 7795
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.534836255682345e-06,
-      "loss": 0.0857,
+      "loss": 0.077,
       "step": 7796
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.532980795992208e-06,
-      "loss": 0.1953,
+      "loss": 0.0747,
       "step": 7797
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.531125336302069e-06,
-      "loss": 0.1384,
+      "loss": 0.0476,
       "step": 7798
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.5292698766119315e-06,
-      "loss": 0.1362,
+      "loss": 0.0109,
       "step": 7799
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.5274144169217925e-06,
-      "loss": 0.1911,
+      "loss": 0.016,
       "step": 7800
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.525558957231655e-06,
-      "loss": 0.1417,
+      "loss": 0.0217,
       "step": 7801
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.523703497541516e-06,
-      "loss": 0.0321,
+      "loss": 0.103,
       "step": 7802
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.521848037851378e-06,
-      "loss": 0.1389,
+      "loss": 0.1464,
       "step": 7803
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.51999257816124e-06,
-      "loss": 0.1392,
+      "loss": 0.0079,
       "step": 7804
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.518137118471102e-06,
-      "loss": 0.0838,
+      "loss": 0.0672,
       "step": 7805
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.516281658780963e-06,
-      "loss": 0.1372,
+      "loss": 0.1102,
       "step": 7806
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.514426199090826e-06,
-      "loss": 0.0323,
+      "loss": 0.1262,
       "step": 7807
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.512570739400687e-06,
-      "loss": 0.1403,
+      "loss": 0.0108,
       "step": 7808
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.510715279710549e-06,
-      "loss": 0.0853,
+      "loss": 0.0523,
       "step": 7809
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.50885982002041e-06,
-      "loss": 0.0866,
+      "loss": 0.0771,
       "step": 7810
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.507004360330272e-06,
-      "loss": 0.1389,
+      "loss": 0.0791,
       "step": 7811
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.505148900640134e-06,
-      "loss": 0.1903,
+      "loss": 0.0452,
       "step": 7812
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.503293440949996e-06,
-      "loss": 0.0865,
+      "loss": 0.0132,
       "step": 7813
     },
     {
       "epoch": 2.17,
       "learning_rate": 5.501437981259857e-06,
-      "loss": 0.0317,
+      "loss": 0.0524,
       "step": 7814
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.49958252156972e-06,
-      "loss": 0.1425,
+      "loss": 0.1758,
       "step": 7815
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.497727061879581e-06,
-      "loss": 0.0875,
+      "loss": 0.0935,
       "step": 7816
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.4958716021894435e-06,
-      "loss": 0.0856,
+      "loss": 0.0759,
       "step": 7817
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.4940161424993045e-06,
-      "loss": 0.0817,
+      "loss": 0.0298,
       "step": 7818
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.492160682809166e-06,
-      "loss": 0.2521,
+      "loss": 0.0546,
       "step": 7819
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.490305223119028e-06,
-      "loss": 0.1929,
+      "loss": 0.0749,
       "step": 7820
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.48844976342889e-06,
-      "loss": 0.1943,
+      "loss": 0.0259,
       "step": 7821
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.486594303738751e-06,
-      "loss": 0.0834,
+      "loss": 0.0538,
       "step": 7822
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.484738844048614e-06,
-      "loss": 0.0866,
+      "loss": 0.0189,
       "step": 7823
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.482883384358475e-06,
-      "loss": 0.1961,
+      "loss": 0.0592,
       "step": 7824
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.4810279246683376e-06,
-      "loss": 0.1971,
+      "loss": 0.0118,
       "step": 7825
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.479172464978199e-06,
-      "loss": 0.0299,
+      "loss": 0.0956,
       "step": 7826
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.4773170052880605e-06,
-      "loss": 0.2476,
+      "loss": 0.0589,
       "step": 7827
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.475461545597922e-06,
-      "loss": 0.0889,
+      "loss": 0.0157,
       "step": 7828
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.473606085907784e-06,
-      "loss": 0.0844,
+      "loss": 0.1762,
       "step": 7829
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.471750626217645e-06,
-      "loss": 0.1366,
+      "loss": 0.0572,
       "step": 7830
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.469895166527508e-06,
-      "loss": 0.1404,
+      "loss": 0.0178,
       "step": 7831
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.468039706837369e-06,
-      "loss": 0.0836,
+      "loss": 0.0597,
       "step": 7832
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.466184247147232e-06,
-      "loss": 0.1946,
+      "loss": 0.1743,
       "step": 7833
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.464328787457093e-06,
-      "loss": 0.2007,
+      "loss": 0.1054,
       "step": 7834
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.4624733277669554e-06,
-      "loss": 0.1384,
+      "loss": 0.0487,
       "step": 7835
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.4606178680768165e-06,
-      "loss": 0.0299,
+      "loss": 0.0654,
       "step": 7836
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.458762408386678e-06,
-      "loss": 0.1385,
+      "loss": 0.0872,
       "step": 7837
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.45690694869654e-06,
-      "loss": 0.1445,
+      "loss": 0.2308,
       "step": 7838
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.455051489006402e-06,
-      "loss": 0.2582,
+      "loss": 0.0171,
       "step": 7839
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.453196029316263e-06,
-      "loss": 0.0845,
+      "loss": 0.0104,
       "step": 7840
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.451340569626126e-06,
-      "loss": 0.1919,
+      "loss": 0.1918,
       "step": 7841
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.449485109935987e-06,
-      "loss": 0.0303,
+      "loss": 0.0494,
       "step": 7842
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.4476296502458495e-06,
-      "loss": 0.0829,
+      "loss": 0.0809,
       "step": 7843
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.4457741905557106e-06,
-      "loss": 0.1377,
+      "loss": 0.1072,
       "step": 7844
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.4439187308655724e-06,
-      "loss": 0.0297,
+      "loss": 0.0493,
       "step": 7845
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.442063271175434e-06,
-      "loss": 0.0841,
+      "loss": 0.0717,
       "step": 7846
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.440207811485296e-06,
-      "loss": 0.0303,
+      "loss": 0.0464,
       "step": 7847
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.438352351795157e-06,
-      "loss": 0.1939,
+      "loss": 0.2631,
       "step": 7848
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.43649689210502e-06,
-      "loss": 0.1914,
+      "loss": 0.051,
       "step": 7849
     },
     {
       "epoch": 2.18,
       "learning_rate": 5.434641432414881e-06,
-      "loss": 0.0834,
+      "loss": 0.1131,
       "step": 7850
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.432785972724744e-06,
-      "loss": 0.1404,
+      "loss": 0.0595,
       "step": 7851
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.430930513034605e-06,
-      "loss": 0.1406,
+      "loss": 0.0635,
       "step": 7852
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.4290750533444666e-06,
-      "loss": 0.0845,
+      "loss": 0.1296,
       "step": 7853
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.4272195936543284e-06,
-      "loss": 0.0287,
+      "loss": 0.0632,
       "step": 7854
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.42536413396419e-06,
-      "loss": 0.0857,
+      "loss": 0.0132,
       "step": 7855
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.423508674274051e-06,
-      "loss": 0.0835,
+      "loss": 0.0589,
       "step": 7856
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.421653214583914e-06,
-      "loss": 0.0803,
+      "loss": 0.0273,
       "step": 7857
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.419797754893775e-06,
-      "loss": 0.3129,
+      "loss": 0.1585,
       "step": 7858
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.417942295203638e-06,
-      "loss": 0.0283,
+      "loss": 0.1233,
       "step": 7859
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.416086835513499e-06,
-      "loss": 0.4201,
+      "loss": 0.0172,
       "step": 7860
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.4142313758233615e-06,
-      "loss": 0.3615,
+      "loss": 0.0147,
       "step": 7861
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.4123759161332225e-06,
-      "loss": 0.1402,
+      "loss": 0.1543,
       "step": 7862
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.410520456443084e-06,
-      "loss": 0.0843,
+      "loss": 0.0568,
       "step": 7863
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.408664996752946e-06,
-      "loss": 0.0833,
+      "loss": 0.0795,
       "step": 7864
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.406809537062808e-06,
-      "loss": 0.0289,
+      "loss": 0.0437,
       "step": 7865
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.404954077372669e-06,
-      "loss": 0.1375,
+      "loss": 0.0153,
       "step": 7866
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.403098617682532e-06,
-      "loss": 0.0297,
+      "loss": 0.0886,
       "step": 7867
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.401243157992393e-06,
-      "loss": 0.1412,
+      "loss": 0.063,
       "step": 7868
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.399387698302256e-06,
-      "loss": 0.0286,
+      "loss": 0.0182,
       "step": 7869
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.397532238612117e-06,
-      "loss": 0.134,
+      "loss": 0.0155,
       "step": 7870
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.3956767789219785e-06,
-      "loss": 0.0842,
+      "loss": 0.0498,
       "step": 7871
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.39382131923184e-06,
-      "loss": 0.3078,
+      "loss": 0.0711,
       "step": 7872
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.391965859541702e-06,
-      "loss": 0.1382,
+      "loss": 0.0652,
       "step": 7873
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.390110399851563e-06,
-      "loss": 0.0835,
+      "loss": 0.0531,
       "step": 7874
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.388254940161426e-06,
-      "loss": 0.0813,
+      "loss": 0.0493,
       "step": 7875
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.386399480471287e-06,
-      "loss": 0.0801,
+      "loss": 0.0935,
       "step": 7876
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.38454402078115e-06,
-      "loss": 0.0828,
+      "loss": 0.0313,
       "step": 7877
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.382688561091011e-06,
-      "loss": 0.1975,
+      "loss": 0.1282,
       "step": 7878
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.380833101400873e-06,
-      "loss": 0.029,
+      "loss": 0.0562,
       "step": 7879
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.3789776417107345e-06,
-      "loss": 0.1969,
+      "loss": 0.1041,
       "step": 7880
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.377122182020596e-06,
-      "loss": 0.0285,
+      "loss": 0.0747,
       "step": 7881
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.375266722330457e-06,
-      "loss": 0.0815,
+      "loss": 0.0583,
       "step": 7882
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.37341126264032e-06,
-      "loss": 0.1435,
+      "loss": 0.1585,
       "step": 7883
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.371555802950181e-06,
-      "loss": 0.0825,
+      "loss": 0.0505,
       "step": 7884
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.369700343260044e-06,
-      "loss": 0.1396,
+      "loss": 0.1353,
       "step": 7885
     },
     {
       "epoch": 2.19,
       "learning_rate": 5.367844883569905e-06,
-      "loss": 0.0276,
+      "loss": 0.0141,
       "step": 7886
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.365989423879767e-06,
-      "loss": 0.1377,
+      "loss": 0.1059,
       "step": 7887
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.364133964189629e-06,
-      "loss": 0.2552,
+      "loss": 0.0879,
       "step": 7888
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.3622785044994905e-06,
-      "loss": 0.0823,
+      "loss": 0.1448,
       "step": 7889
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.3604230448093515e-06,
-      "loss": 0.082,
+      "loss": 0.1361,
       "step": 7890
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.358567585119214e-06,
-      "loss": 0.3052,
+      "loss": 0.0213,
       "step": 7891
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.356712125429075e-06,
-      "loss": 0.1363,
+      "loss": 0.0974,
       "step": 7892
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.354856665738938e-06,
-      "loss": 0.1406,
+      "loss": 0.0356,
       "step": 7893
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.353001206048799e-06,
-      "loss": 0.1424,
+      "loss": 0.0151,
       "step": 7894
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.35114574635866e-06,
-      "loss": 0.0845,
+      "loss": 0.081,
       "step": 7895
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.349290286668523e-06,
-      "loss": 0.1431,
+      "loss": 0.1386,
       "step": 7896
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.347434826978384e-06,
-      "loss": 0.1414,
+      "loss": 0.1102,
       "step": 7897
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.3455793672882465e-06,
-      "loss": 0.083,
+      "loss": 0.1838,
       "step": 7898
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.3437239075981075e-06,
-      "loss": 0.1935,
+      "loss": 0.0906,
       "step": 7899
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.341868447907969e-06,
-      "loss": 0.0863,
+      "loss": 0.0625,
       "step": 7900
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.340012988217831e-06,
-      "loss": 0.1959,
+      "loss": 0.0302,
       "step": 7901
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.338157528527693e-06,
-      "loss": 0.086,
+      "loss": 0.0786,
       "step": 7902
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.336302068837554e-06,
-      "loss": 0.2026,
+      "loss": 0.1243,
       "step": 7903
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.334446609147417e-06,
-      "loss": 0.0846,
+      "loss": 0.0219,
       "step": 7904
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.332591149457278e-06,
-      "loss": 0.3633,
+      "loss": 0.0463,
       "step": 7905
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.330735689767141e-06,
-      "loss": 0.141,
+      "loss": 0.036,
       "step": 7906
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.328880230077002e-06,
-      "loss": 0.1943,
+      "loss": 0.0215,
       "step": 7907
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.3270247703868635e-06,
-      "loss": 0.1398,
+      "loss": 0.0259,
       "step": 7908
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.325169310696725e-06,
-      "loss": 0.0832,
+      "loss": 0.1394,
       "step": 7909
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.323313851006587e-06,
-      "loss": 0.1992,
+      "loss": 0.1714,
       "step": 7910
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.321458391316448e-06,
-      "loss": 0.0888,
+      "loss": 0.064,
       "step": 7911
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.319602931626311e-06,
-      "loss": 0.0803,
+      "loss": 0.0167,
       "step": 7912
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.317747471936172e-06,
-      "loss": 0.1417,
+      "loss": 0.0575,
       "step": 7913
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.315892012246035e-06,
-      "loss": 0.0843,
+      "loss": 0.045,
       "step": 7914
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.314036552555896e-06,
-      "loss": 0.198,
+      "loss": 0.0209,
       "step": 7915
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.312181092865758e-06,
-      "loss": 0.3579,
+      "loss": 0.0188,
       "step": 7916
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.3103256331756195e-06,
-      "loss": 0.0299,
+      "loss": 0.0725,
       "step": 7917
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.308470173485481e-06,
-      "loss": 0.0835,
+      "loss": 0.0829,
       "step": 7918
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.306614713795342e-06,
-      "loss": 0.3049,
+      "loss": 0.0509,
       "step": 7919
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.304759254105205e-06,
-      "loss": 0.1926,
+      "loss": 0.011,
       "step": 7920
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.302903794415066e-06,
-      "loss": 0.1385,
+      "loss": 0.0227,
       "step": 7921
     },
     {
       "epoch": 2.2,
       "learning_rate": 5.301048334724929e-06,
-      "loss": 0.0855,
+      "loss": 0.0441,
       "step": 7922
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.29919287503479e-06,
-      "loss": 0.1939,
+      "loss": 0.0566,
       "step": 7923
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.297337415344653e-06,
-      "loss": 0.1933,
+      "loss": 0.0644,
       "step": 7924
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.295481955654514e-06,
-      "loss": 0.248,
+      "loss": 0.1557,
       "step": 7925
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.2936264959643755e-06,
-      "loss": 0.2478,
+      "loss": 0.1003,
       "step": 7926
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.291771036274237e-06,
-      "loss": 0.0874,
+      "loss": 0.0965,
       "step": 7927
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.289915576584099e-06,
-      "loss": 0.0856,
+      "loss": 0.1467,
       "step": 7928
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.28806011689396e-06,
-      "loss": 0.0853,
+      "loss": 0.1258,
       "step": 7929
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.286204657203823e-06,
-      "loss": 0.1396,
+      "loss": 0.0478,
       "step": 7930
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.284349197513684e-06,
-      "loss": 0.0838,
+      "loss": 0.0087,
       "step": 7931
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.282493737823547e-06,
-      "loss": 0.1873,
+      "loss": 0.0558,
       "step": 7932
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.280638278133408e-06,
-      "loss": 0.1343,
+      "loss": 0.044,
       "step": 7933
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.27878281844327e-06,
-      "loss": 0.0334,
+      "loss": 0.0534,
       "step": 7934
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.2769273587531315e-06,
-      "loss": 0.0849,
+      "loss": 0.0173,
       "step": 7935
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.275071899062993e-06,
-      "loss": 0.1379,
+      "loss": 0.015,
       "step": 7936
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.273216439372854e-06,
-      "loss": 0.032,
+      "loss": 0.1233,
       "step": 7937
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.271360979682717e-06,
-      "loss": 0.0831,
+      "loss": 0.0916,
       "step": 7938
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.269505519992578e-06,
-      "loss": 0.1441,
+      "loss": 0.0216,
       "step": 7939
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.267650060302441e-06,
-      "loss": 0.0875,
+      "loss": 0.0168,
       "step": 7940
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.265794600612302e-06,
-      "loss": 0.1421,
+      "loss": 0.0608,
       "step": 7941
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.263939140922164e-06,
-      "loss": 0.0313,
+      "loss": 0.2016,
       "step": 7942
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.262083681232026e-06,
-      "loss": 0.1346,
+      "loss": 0.0812,
       "step": 7943
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.2602282215418875e-06,
-      "loss": 0.1405,
+      "loss": 0.0465,
       "step": 7944
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.2583727618517485e-06,
-      "loss": 0.1939,
+      "loss": 0.0519,
       "step": 7945
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.256517302161611e-06,
-      "loss": 0.0848,
+      "loss": 0.1847,
       "step": 7946
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.254661842471472e-06,
-      "loss": 0.1398,
+      "loss": 0.0653,
       "step": 7947
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.252806382781335e-06,
-      "loss": 0.1388,
+      "loss": 0.0666,
       "step": 7948
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.250950923091196e-06,
-      "loss": 0.356,
+      "loss": 0.0945,
       "step": 7949
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.249095463401058e-06,
-      "loss": 0.0314,
+      "loss": 0.0994,
       "step": 7950
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.24724000371092e-06,
-      "loss": 0.1401,
+      "loss": 0.0435,
       "step": 7951
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.245384544020782e-06,
-      "loss": 0.0295,
+      "loss": 0.1106,
       "step": 7952
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.243529084330643e-06,
-      "loss": 0.253,
+      "loss": 0.057,
       "step": 7953
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.241673624640505e-06,
-      "loss": 0.0852,
+      "loss": 0.1351,
       "step": 7954
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.239818164950366e-06,
-      "loss": 0.1395,
+      "loss": 0.0195,
       "step": 7955
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.237962705260229e-06,
-      "loss": 0.2463,
+      "loss": 0.0566,
       "step": 7956
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.23610724557009e-06,
-      "loss": 0.0867,
+      "loss": 0.1069,
       "step": 7957
     },
     {
       "epoch": 2.21,
       "learning_rate": 5.234251785879953e-06,
-      "loss": 0.1406,
+      "loss": 0.0181,
       "step": 7958
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.232396326189814e-06,
-      "loss": 0.1919,
+      "loss": 0.069,
       "step": 7959
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.230540866499676e-06,
-      "loss": 0.1991,
+      "loss": 0.0257,
       "step": 7960
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.2286854068095376e-06,
-      "loss": 0.0307,
+      "loss": 0.1411,
       "step": 7961
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.2268299471193994e-06,
-      "loss": 0.0847,
+      "loss": 0.0886,
       "step": 7962
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.2249744874292605e-06,
-      "loss": 0.14,
+      "loss": 0.0217,
       "step": 7963
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.223119027739123e-06,
-      "loss": 0.1384,
+      "loss": 0.0457,
       "step": 7964
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.221263568048984e-06,
-      "loss": 0.2469,
+      "loss": 0.0148,
       "step": 7965
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.219408108358847e-06,
-      "loss": 0.1377,
+      "loss": 0.0733,
       "step": 7966
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.217552648668708e-06,
-      "loss": 0.1429,
+      "loss": 0.0779,
       "step": 7967
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.21569718897857e-06,
-      "loss": 0.0311,
+      "loss": 0.1144,
       "step": 7968
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.213841729288432e-06,
-      "loss": 0.0866,
+      "loss": 0.049,
       "step": 7969
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.2119862695982936e-06,
-      "loss": 0.19,
+      "loss": 0.028,
       "step": 7970
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.210130809908155e-06,
-      "loss": 0.1383,
+      "loss": 0.1009,
       "step": 7971
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.208275350218017e-06,
-      "loss": 0.1399,
+      "loss": 0.0708,
       "step": 7972
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.206419890527878e-06,
-      "loss": 0.1954,
+      "loss": 0.1358,
       "step": 7973
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.204564430837741e-06,
-      "loss": 0.0312,
+      "loss": 0.1277,
       "step": 7974
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.202708971147602e-06,
-      "loss": 0.1944,
+      "loss": 0.01,
       "step": 7975
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.200853511457464e-06,
-      "loss": 0.1352,
+      "loss": 0.151,
       "step": 7976
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.198998051767326e-06,
-      "loss": 0.189,
+      "loss": 0.052,
       "step": 7977
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.197142592077188e-06,
-      "loss": 0.0295,
+      "loss": 0.0238,
       "step": 7978
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.195287132387049e-06,
-      "loss": 0.1931,
+      "loss": 0.1094,
       "step": 7979
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.193431672696911e-06,
-      "loss": 0.1369,
+      "loss": 0.1259,
       "step": 7980
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.1915762130067724e-06,
-      "loss": 0.0303,
+      "loss": 0.0233,
       "step": 7981
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.189720753316635e-06,
-      "loss": 0.0831,
+      "loss": 0.0215,
       "step": 7982
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.187865293626496e-06,
-      "loss": 0.1407,
+      "loss": 0.019,
       "step": 7983
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.186009833936359e-06,
-      "loss": 0.0852,
+      "loss": 0.0592,
       "step": 7984
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.18415437424622e-06,
-      "loss": 0.0294,
+      "loss": 0.0679,
       "step": 7985
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.182298914556082e-06,
-      "loss": 0.3015,
+      "loss": 0.0202,
       "step": 7986
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.180443454865944e-06,
-      "loss": 0.1367,
+      "loss": 0.0718,
       "step": 7987
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.1785879951758055e-06,
-      "loss": 0.0852,
+      "loss": 0.1037,
       "step": 7988
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.1767325354856666e-06,
-      "loss": 0.141,
+      "loss": 0.0631,
       "step": 7989
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.174877075795529e-06,
-      "loss": 0.1383,
+      "loss": 0.0622,
       "step": 7990
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.17302161610539e-06,
-      "loss": 0.1376,
+      "loss": 0.0755,
       "step": 7991
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.171166156415253e-06,
-      "loss": 0.1414,
+      "loss": 0.0462,
       "step": 7992
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.169310696725114e-06,
-      "loss": 0.0836,
+      "loss": 0.1065,
       "step": 7993
     },
     {
       "epoch": 2.22,
       "learning_rate": 5.167455237034976e-06,
-      "loss": 0.0848,
+      "loss": 0.1331,
       "step": 7994
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.165599777344838e-06,
-      "loss": 0.0866,
+      "loss": 0.1186,
       "step": 7995
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.1637443176547e-06,
-      "loss": 0.196,
+      "loss": 0.0448,
       "step": 7996
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.161888857964561e-06,
-      "loss": 0.1917,
+      "loss": 0.0133,
       "step": 7997
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.160033398274423e-06,
-      "loss": 0.1392,
+      "loss": 0.1591,
       "step": 7998
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.158177938584284e-06,
-      "loss": 0.2421,
+      "loss": 0.1006,
       "step": 7999
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.156322478894147e-06,
-      "loss": 0.3616,
+      "loss": 0.0965,
       "step": 8000
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.154467019204008e-06,
-      "loss": 0.0862,
+      "loss": 0.0348,
       "step": 8001
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.15261155951387e-06,
-      "loss": 0.1947,
+      "loss": 0.0615,
       "step": 8002
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.150756099823732e-06,
-      "loss": 0.086,
+      "loss": 0.0125,
       "step": 8003
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.148900640133594e-06,
-      "loss": 0.2493,
+      "loss": 0.1131,
       "step": 8004
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.147045180443455e-06,
-      "loss": 0.3517,
+      "loss": 0.143,
       "step": 8005
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.1451897207533175e-06,
-      "loss": 0.0862,
+      "loss": 0.0915,
       "step": 8006
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.1433342610631785e-06,
-      "loss": 0.2504,
+      "loss": 0.0636,
       "step": 8007
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.141478801373041e-06,
-      "loss": 0.0882,
+      "loss": 0.0369,
       "step": 8008
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.139623341682902e-06,
-      "loss": 0.1845,
+      "loss": 0.228,
       "step": 8009
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.137767881992764e-06,
-      "loss": 0.1379,
+      "loss": 0.1714,
       "step": 8010
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.135912422302626e-06,
-      "loss": 0.0887,
+      "loss": 0.0533,
       "step": 8011
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.134056962612488e-06,
-      "loss": 0.1987,
+      "loss": 0.0169,
       "step": 8012
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.132201502922349e-06,
-      "loss": 0.0852,
+      "loss": 0.0225,
       "step": 8013
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.130346043232212e-06,
-      "loss": 0.2512,
+      "loss": 0.116,
       "step": 8014
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.128490583542073e-06,
-      "loss": 0.0347,
+      "loss": 0.099,
       "step": 8015
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.126635123851935e-06,
-      "loss": 0.1955,
+      "loss": 0.1355,
       "step": 8016
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.124779664161796e-06,
-      "loss": 0.1942,
+      "loss": 0.1239,
       "step": 8017
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.122924204471659e-06,
-      "loss": 0.2395,
+      "loss": 0.0686,
       "step": 8018
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.12106874478152e-06,
-      "loss": 0.1359,
+      "loss": 0.0276,
       "step": 8019
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.119213285091382e-06,
-      "loss": 0.0875,
+      "loss": 0.0657,
       "step": 8020
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.117357825401244e-06,
-      "loss": 0.0354,
+      "loss": 0.1227,
       "step": 8021
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.115502365711106e-06,
-      "loss": 0.1901,
+      "loss": 0.1457,
       "step": 8022
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.113646906020967e-06,
-      "loss": 0.0898,
+      "loss": 0.0591,
       "step": 8023
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.1117914463308295e-06,
-      "loss": 0.2423,
+      "loss": 0.1092,
       "step": 8024
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.1099359866406905e-06,
-      "loss": 0.1378,
+      "loss": 0.0767,
       "step": 8025
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.108080526950553e-06,
-      "loss": 0.2453,
+      "loss": 0.1015,
       "step": 8026
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.106225067260414e-06,
-      "loss": 0.1388,
+      "loss": 0.079,
       "step": 8027
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.104369607570276e-06,
-      "loss": 0.2432,
+      "loss": 0.0191,
       "step": 8028
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.102514147880138e-06,
-      "loss": 0.143,
+      "loss": 0.1379,
       "step": 8029
     },
     {
       "epoch": 2.23,
       "learning_rate": 5.10065868819e-06,
-      "loss": 0.1445,
+      "loss": 0.1337,
       "step": 8030
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.098803228499861e-06,
-      "loss": 0.1388,
+      "loss": 0.082,
       "step": 8031
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.096947768809724e-06,
-      "loss": 0.0372,
+      "loss": 0.1002,
       "step": 8032
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.095092309119585e-06,
-      "loss": 0.2437,
+      "loss": 0.0218,
       "step": 8033
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.093236849429447e-06,
-      "loss": 0.1373,
+      "loss": 0.032,
       "step": 8034
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.091381389739308e-06,
-      "loss": 0.0866,
+      "loss": 0.0703,
       "step": 8035
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.08952593004917e-06,
-      "loss": 0.0369,
+      "loss": 0.0879,
       "step": 8036
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.087670470359032e-06,
-      "loss": 0.0363,
+      "loss": 0.0559,
       "step": 8037
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.085815010668894e-06,
-      "loss": 0.1895,
+      "loss": 0.0498,
       "step": 8038
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.083959550978755e-06,
-      "loss": 0.0347,
+      "loss": 0.0964,
       "step": 8039
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.082104091288618e-06,
-      "loss": 0.1935,
+      "loss": 0.0307,
       "step": 8040
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.080248631598479e-06,
-      "loss": 0.1433,
+      "loss": 0.0182,
       "step": 8041
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.0783931719083415e-06,
-      "loss": 0.0891,
+      "loss": 0.0757,
       "step": 8042
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.0765377122182025e-06,
-      "loss": 0.1389,
+      "loss": 0.0972,
       "step": 8043
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.074682252528065e-06,
-      "loss": 0.1441,
+      "loss": 0.0212,
       "step": 8044
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.072826792837926e-06,
-      "loss": 0.2461,
+      "loss": 0.0472,
       "step": 8045
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.070971333147788e-06,
-      "loss": 0.0346,
+      "loss": 0.0518,
       "step": 8046
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.06911587345765e-06,
-      "loss": 0.1902,
+      "loss": 0.0797,
       "step": 8047
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.067260413767511e-06,
-      "loss": 0.142,
+      "loss": 0.0687,
       "step": 8048
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.065404954077373e-06,
-      "loss": 0.1903,
+      "loss": 0.0591,
       "step": 8049
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.063549494387235e-06,
-      "loss": 0.1402,
+      "loss": 0.0134,
       "step": 8050
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.061694034697097e-06,
-      "loss": 0.0843,
+      "loss": 0.0179,
       "step": 8051
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.059838575006958e-06,
-      "loss": 0.1319,
+      "loss": 0.0284,
       "step": 8052
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.05798311531682e-06,
-      "loss": 0.0333,
+      "loss": 0.0652,
       "step": 8053
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.056127655626681e-06,
-      "loss": 0.2371,
+      "loss": 0.1776,
       "step": 8054
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.054272195936544e-06,
-      "loss": 0.1937,
+      "loss": 0.1211,
       "step": 8055
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.052416736246405e-06,
-      "loss": 0.4052,
+      "loss": 0.0789,
       "step": 8056
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.050561276556267e-06,
-      "loss": 0.137,
+      "loss": 0.1459,
       "step": 8057
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.048705816866129e-06,
-      "loss": 0.0337,
+      "loss": 0.0482,
       "step": 8058
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.046850357175991e-06,
-      "loss": 0.1407,
+      "loss": 0.0646,
       "step": 8059
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.044994897485852e-06,
-      "loss": 0.0337,
+      "loss": 0.0504,
       "step": 8060
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.0431394377957145e-06,
-      "loss": 0.0884,
+      "loss": 0.0129,
       "step": 8061
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.0412839781055755e-06,
-      "loss": 0.0871,
+      "loss": 0.0546,
       "step": 8062
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.039428518415438e-06,
-      "loss": 0.1385,
+      "loss": 0.1219,
       "step": 8063
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.037573058725299e-06,
-      "loss": 0.2495,
+      "loss": 0.0248,
       "step": 8064
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.035717599035161e-06,
-      "loss": 0.2434,
+      "loss": 0.0109,
       "step": 8065
     },
     {
       "epoch": 2.24,
       "learning_rate": 5.033862139345023e-06,
-      "loss": 0.1905,
+      "loss": 0.1165,
       "step": 8066
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.032006679654885e-06,
-      "loss": 0.0337,
+      "loss": 0.0913,
       "step": 8067
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.030151219964746e-06,
-      "loss": 0.0343,
+      "loss": 0.0205,
       "step": 8068
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.028295760274609e-06,
-      "loss": 0.1924,
+      "loss": 0.0919,
       "step": 8069
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.02644030058447e-06,
-      "loss": 0.3503,
+      "loss": 0.0506,
       "step": 8070
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.024584840894332e-06,
-      "loss": 0.244,
+      "loss": 0.1657,
       "step": 8071
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.022729381204193e-06,
-      "loss": 0.0855,
+      "loss": 0.0512,
       "step": 8072
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.020873921514055e-06,
-      "loss": 0.1387,
+      "loss": 0.0638,
       "step": 8073
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.019018461823917e-06,
-      "loss": 0.0862,
+      "loss": 0.1427,
       "step": 8074
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.017163002133779e-06,
-      "loss": 0.1359,
+      "loss": 0.1413,
       "step": 8075
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.01530754244364e-06,
-      "loss": 0.1362,
+      "loss": 0.0122,
       "step": 8076
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.013452082753503e-06,
-      "loss": 0.1896,
+      "loss": 0.039,
       "step": 8077
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.011596623063364e-06,
-      "loss": 0.241,
+      "loss": 0.1344,
       "step": 8078
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.0097411633732264e-06,
-      "loss": 0.1388,
+      "loss": 0.1436,
       "step": 8079
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.0078857036830875e-06,
-      "loss": 0.0861,
+      "loss": 0.1317,
       "step": 8080
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.00603024399295e-06,
-      "loss": 0.2488,
+      "loss": 0.1008,
       "step": 8081
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.004174784302811e-06,
-      "loss": 0.0844,
+      "loss": 0.1359,
       "step": 8082
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.002319324612673e-06,
-      "loss": 0.2899,
+      "loss": 0.0153,
       "step": 8083
     },
     {
       "epoch": 2.25,
       "learning_rate": 5.000463864922535e-06,
-      "loss": 0.1423,
+      "loss": 0.0555,
       "step": 8084
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.998608405232397e-06,
-      "loss": 0.0886,
+      "loss": 0.0888,
       "step": 8085
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.996752945542259e-06,
-      "loss": 0.2938,
+      "loss": 0.1275,
       "step": 8086
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.9948974858521206e-06,
-      "loss": 0.1396,
+      "loss": 0.1428,
       "step": 8087
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.9930420261619824e-06,
-      "loss": 0.1416,
+      "loss": 0.0301,
       "step": 8088
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.991186566471844e-06,
-      "loss": 0.1414,
+      "loss": 0.1142,
       "step": 8089
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.989331106781705e-06,
-      "loss": 0.2439,
+      "loss": 0.1317,
       "step": 8090
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.987475647091567e-06,
-      "loss": 0.1404,
+      "loss": 0.0388,
       "step": 8091
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.985620187401429e-06,
-      "loss": 0.2432,
+      "loss": 0.0996,
       "step": 8092
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.983764727711291e-06,
-      "loss": 0.0896,
+      "loss": 0.0537,
       "step": 8093
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.981909268021153e-06,
-      "loss": 0.0379,
+      "loss": 0.0389,
       "step": 8094
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.980053808331015e-06,
-      "loss": 0.0366,
+      "loss": 0.0319,
       "step": 8095
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.9781983486408765e-06,
-      "loss": 0.0363,
+      "loss": 0.0967,
       "step": 8096
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.976342888950738e-06,
-      "loss": 0.1901,
+      "loss": 0.0526,
       "step": 8097
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.9744874292606e-06,
-      "loss": 0.0875,
+      "loss": 0.0245,
       "step": 8098
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.972631969570461e-06,
-      "loss": 0.0874,
+      "loss": 0.0337,
       "step": 8099
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.970776509880323e-06,
-      "loss": 0.0902,
+      "loss": 0.0312,
       "step": 8100
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.968921050190185e-06,
-      "loss": 0.1921,
+      "loss": 0.0933,
       "step": 8101
     },
     {
       "epoch": 2.25,
       "learning_rate": 4.967065590500047e-06,
-      "loss": 0.0345,
+      "loss": 0.0584,
       "step": 8102
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.965210130809909e-06,
-      "loss": 0.1912,
+      "loss": 0.0528,
       "step": 8103
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.963354671119771e-06,
-      "loss": 0.1372,
+      "loss": 0.0705,
       "step": 8104
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.9614992114296325e-06,
-      "loss": 0.1353,
+      "loss": 0.0934,
       "step": 8105
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.959643751739494e-06,
-      "loss": 0.0839,
+      "loss": 0.0915,
       "step": 8106
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.957788292049356e-06,
-      "loss": 0.0346,
+      "loss": 0.0159,
       "step": 8107
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.955932832359217e-06,
-      "loss": 0.1926,
+      "loss": 0.0986,
       "step": 8108
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.954077372669079e-06,
-      "loss": 0.2459,
+      "loss": 0.0212,
       "step": 8109
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.952221912978941e-06,
-      "loss": 0.0321,
+      "loss": 0.0188,
       "step": 8110
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.950366453288803e-06,
-      "loss": 0.0852,
+      "loss": 0.0173,
       "step": 8111
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.948510993598665e-06,
-      "loss": 0.1433,
+      "loss": 0.1321,
       "step": 8112
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.946655533908527e-06,
-      "loss": 0.1432,
+      "loss": 0.0955,
       "step": 8113
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.9448000742183885e-06,
-      "loss": 0.5123,
+      "loss": 0.0126,
       "step": 8114
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.94294461452825e-06,
-      "loss": 0.2444,
+      "loss": 0.0961,
       "step": 8115
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.941089154838111e-06,
-      "loss": 0.0324,
+      "loss": 0.0199,
       "step": 8116
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.939233695147973e-06,
-      "loss": 0.1975,
+      "loss": 0.0431,
       "step": 8117
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.937378235457835e-06,
-      "loss": 0.1384,
+      "loss": 0.087,
       "step": 8118
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.935522775767697e-06,
-      "loss": 0.1961,
+      "loss": 0.0536,
       "step": 8119
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.933667316077559e-06,
-      "loss": 0.0844,
+      "loss": 0.0175,
       "step": 8120
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.931811856387421e-06,
-      "loss": 0.19,
+      "loss": 0.0158,
       "step": 8121
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.929956396697283e-06,
-      "loss": 0.1894,
+      "loss": 0.0619,
       "step": 8122
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.928100937007144e-06,
-      "loss": 0.1951,
+      "loss": 0.1057,
       "step": 8123
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.9262454773170055e-06,
-      "loss": 0.1393,
+      "loss": 0.1155,
       "step": 8124
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.924390017626867e-06,
-      "loss": 0.1864,
+      "loss": 0.0511,
       "step": 8125
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.922534557936729e-06,
-      "loss": 0.0859,
+      "loss": 0.0384,
       "step": 8126
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.920679098246591e-06,
-      "loss": 0.1368,
+      "loss": 0.0072,
       "step": 8127
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.918823638556452e-06,
-      "loss": 0.1406,
+      "loss": 0.0193,
       "step": 8128
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.916968178866314e-06,
-      "loss": 0.1872,
+      "loss": 0.044,
       "step": 8129
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.915112719176176e-06,
-      "loss": 0.1927,
+      "loss": 0.1141,
       "step": 8130
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.913257259486038e-06,
-      "loss": 0.142,
+      "loss": 0.0081,
       "step": 8131
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.9114017997959e-06,
-      "loss": 0.0339,
+      "loss": 0.122,
       "step": 8132
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.9095463401057615e-06,
-      "loss": 0.1878,
+      "loss": 0.0107,
       "step": 8133
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.907690880415623e-06,
-      "loss": 0.034,
+      "loss": 0.092,
       "step": 8134
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.905835420725485e-06,
-      "loss": 0.1405,
+      "loss": 0.1007,
       "step": 8135
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.903979961035346e-06,
-      "loss": 0.0849,
+      "loss": 0.0185,
       "step": 8136
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.902124501345208e-06,
-      "loss": 0.0864,
+      "loss": 0.0634,
       "step": 8137
     },
     {
       "epoch": 2.26,
       "learning_rate": 4.90026904165507e-06,
-      "loss": 0.0339,
+      "loss": 0.0643,
       "step": 8138
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.898413581964932e-06,
-      "loss": 0.1363,
+      "loss": 0.0339,
       "step": 8139
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.896558122274794e-06,
-      "loss": 0.1369,
+      "loss": 0.0316,
       "step": 8140
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.894702662584656e-06,
-      "loss": 0.138,
+      "loss": 0.0143,
       "step": 8141
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.8928472028945175e-06,
-      "loss": 0.0333,
+      "loss": 0.1164,
       "step": 8142
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.890991743204379e-06,
-      "loss": 0.1349,
+      "loss": 0.043,
       "step": 8143
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.889136283514241e-06,
-      "loss": 0.1356,
+      "loss": 0.0437,
       "step": 8144
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.887280823824102e-06,
-      "loss": 0.1369,
+      "loss": 0.1581,
       "step": 8145
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.885425364133964e-06,
-      "loss": 0.3028,
+      "loss": 0.1056,
       "step": 8146
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.883569904443826e-06,
-      "loss": 0.2986,
+      "loss": 0.0538,
       "step": 8147
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.881714444753688e-06,
-      "loss": 0.1945,
+      "loss": 0.0681,
       "step": 8148
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.87985898506355e-06,
-      "loss": 0.1374,
+      "loss": 0.0699,
       "step": 8149
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.878003525373412e-06,
-      "loss": 0.2505,
+      "loss": 0.0298,
       "step": 8150
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.8761480656832735e-06,
-      "loss": 0.1899,
+      "loss": 0.0802,
       "step": 8151
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.874292605993135e-06,
-      "loss": 0.0327,
+      "loss": 0.136,
       "step": 8152
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.872437146302996e-06,
-      "loss": 0.2466,
+      "loss": 0.07,
       "step": 8153
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.870581686612858e-06,
-      "loss": 0.1397,
+      "loss": 0.0208,
       "step": 8154
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.86872622692272e-06,
-      "loss": 0.0861,
+      "loss": 0.0836,
       "step": 8155
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.866870767232582e-06,
-      "loss": 0.1394,
+      "loss": 0.0211,
       "step": 8156
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.865015307542444e-06,
-      "loss": 0.1906,
+      "loss": 0.1365,
       "step": 8157
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.863159847852306e-06,
-      "loss": 0.0335,
+      "loss": 0.0498,
       "step": 8158
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.861304388162168e-06,
-      "loss": 0.1916,
+      "loss": 0.1333,
       "step": 8159
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.8594489284720295e-06,
-      "loss": 0.2454,
+      "loss": 0.0384,
       "step": 8160
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.857593468781891e-06,
-      "loss": 0.1401,
+      "loss": 0.0213,
       "step": 8161
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.855738009091752e-06,
-      "loss": 0.2437,
+      "loss": 0.0206,
       "step": 8162
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.853882549401614e-06,
-      "loss": 0.1952,
+      "loss": 0.1091,
       "step": 8163
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.852027089711476e-06,
-      "loss": 0.134,
+      "loss": 0.0115,
       "step": 8164
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.850171630021338e-06,
-      "loss": 0.0349,
+      "loss": 0.067,
       "step": 8165
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.8483161703312e-06,
-      "loss": 0.1369,
+      "loss": 0.0521,
       "step": 8166
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.846460710641062e-06,
-      "loss": 0.0331,
+      "loss": 0.0138,
       "step": 8167
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.844605250950924e-06,
-      "loss": 0.1927,
+      "loss": 0.0177,
       "step": 8168
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.8427497912607855e-06,
-      "loss": 0.3543,
+      "loss": 0.0878,
       "step": 8169
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.840894331570647e-06,
-      "loss": 0.1401,
+      "loss": 0.1506,
       "step": 8170
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.839038871880508e-06,
-      "loss": 0.1393,
+      "loss": 0.0191,
       "step": 8171
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.83718341219037e-06,
-      "loss": 0.0343,
+      "loss": 0.0558,
       "step": 8172
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.835327952500232e-06,
-      "loss": 0.1929,
+      "loss": 0.0109,
       "step": 8173
     },
     {
       "epoch": 2.27,
       "learning_rate": 4.833472492810094e-06,
-      "loss": 0.0879,
+      "loss": 0.1271,
       "step": 8174
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.831617033119956e-06,
-      "loss": 0.0842,
+      "loss": 0.1217,
       "step": 8175
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.829761573429818e-06,
-      "loss": 0.1872,
+      "loss": 0.1065,
       "step": 8176
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.82790611373968e-06,
-      "loss": 0.0849,
+      "loss": 0.074,
       "step": 8177
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.8260506540495415e-06,
-      "loss": 0.1917,
+      "loss": 0.0251,
       "step": 8178
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.8241951943594025e-06,
-      "loss": 0.1387,
+      "loss": 0.1245,
       "step": 8179
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.822339734669264e-06,
-      "loss": 0.0881,
+      "loss": 0.1154,
       "step": 8180
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.820484274979126e-06,
-      "loss": 0.0878,
+      "loss": 0.1035,
       "step": 8181
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.818628815288988e-06,
-      "loss": 0.0344,
+      "loss": 0.0748,
       "step": 8182
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.81677335559885e-06,
-      "loss": 0.141,
+      "loss": 0.1299,
       "step": 8183
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.814917895908712e-06,
-      "loss": 0.0346,
+      "loss": 0.0324,
       "step": 8184
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.813062436218574e-06,
-      "loss": 0.1896,
+      "loss": 0.0353,
       "step": 8185
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.8112069765284356e-06,
-      "loss": 0.1918,
+      "loss": 0.0158,
       "step": 8186
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.8093515168382974e-06,
-      "loss": 0.1911,
+      "loss": 0.1371,
       "step": 8187
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.8074960571481585e-06,
-      "loss": 0.1374,
+      "loss": 0.046,
       "step": 8188
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.80564059745802e-06,
-      "loss": 0.2974,
+      "loss": 0.1146,
       "step": 8189
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.803785137767882e-06,
-      "loss": 0.2444,
+      "loss": 0.0166,
       "step": 8190
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.801929678077744e-06,
-      "loss": 0.2439,
+      "loss": 0.0094,
       "step": 8191
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.800074218387606e-06,
-      "loss": 0.0859,
+      "loss": 0.1153,
       "step": 8192
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.798218758697468e-06,
-      "loss": 0.1897,
+      "loss": 0.0737,
       "step": 8193
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.79636329900733e-06,
-      "loss": 0.0874,
+      "loss": 0.1239,
       "step": 8194
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.7945078393171916e-06,
-      "loss": 0.192,
+      "loss": 0.0806,
       "step": 8195
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.792652379627053e-06,
-      "loss": 0.0847,
+      "loss": 0.0195,
       "step": 8196
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.7907969199369145e-06,
-      "loss": 0.0875,
+      "loss": 0.1066,
       "step": 8197
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.788941460246776e-06,
-      "loss": 0.1398,
+      "loss": 0.128,
       "step": 8198
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.787086000556638e-06,
-      "loss": 0.1904,
+      "loss": 0.1504,
       "step": 8199
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.7852305408665e-06,
-      "loss": 0.1396,
+      "loss": 0.1606,
       "step": 8200
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.783375081176362e-06,
-      "loss": 0.1388,
+      "loss": 0.019,
       "step": 8201
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.781519621486224e-06,
-      "loss": 0.0851,
+      "loss": 0.0141,
       "step": 8202
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.779664161796086e-06,
-      "loss": 0.1384,
+      "loss": 0.0927,
       "step": 8203
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.7778087021059476e-06,
-      "loss": 0.0336,
+      "loss": 0.0688,
       "step": 8204
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.7759532424158086e-06,
-      "loss": 0.0876,
+      "loss": 0.1024,
       "step": 8205
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.7740977827256704e-06,
-      "loss": 0.1357,
+      "loss": 0.1315,
       "step": 8206
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.772242323035532e-06,
-      "loss": 0.1383,
+      "loss": 0.0706,
       "step": 8207
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.770386863345394e-06,
-      "loss": 0.0863,
+      "loss": 0.0425,
       "step": 8208
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.768531403655256e-06,
-      "loss": 0.1911,
+      "loss": 0.0198,
       "step": 8209
     },
     {
       "epoch": 2.28,
       "learning_rate": 4.766675943965118e-06,
-      "loss": 0.1354,
+      "loss": 0.0224,
       "step": 8210
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.76482048427498e-06,
-      "loss": 0.2428,
+      "loss": 0.0148,
       "step": 8211
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.762965024584842e-06,
-      "loss": 0.1406,
+      "loss": 0.0349,
       "step": 8212
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.761109564894703e-06,
-      "loss": 0.0858,
+      "loss": 0.0128,
       "step": 8213
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.7592541052045646e-06,
-      "loss": 0.0325,
+      "loss": 0.1036,
       "step": 8214
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.7573986455144264e-06,
-      "loss": 0.0835,
+      "loss": 0.0425,
       "step": 8215
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.755543185824288e-06,
-      "loss": 0.1401,
+      "loss": 0.1082,
       "step": 8216
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.75368772613415e-06,
-      "loss": 0.1382,
+      "loss": 0.108,
       "step": 8217
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.751832266444012e-06,
-      "loss": 0.3622,
+      "loss": 0.0612,
       "step": 8218
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.749976806753874e-06,
-      "loss": 0.0872,
+      "loss": 0.126,
       "step": 8219
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.748121347063736e-06,
-      "loss": 0.0843,
+      "loss": 0.0177,
       "step": 8220
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.746265887373598e-06,
-      "loss": 0.0861,
+      "loss": 0.0452,
       "step": 8221
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.744410427683459e-06,
-      "loss": 0.139,
+      "loss": 0.0468,
       "step": 8222
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.7425549679933206e-06,
-      "loss": 0.4072,
+      "loss": 0.0585,
       "step": 8223
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.740699508303182e-06,
-      "loss": 0.2986,
+      "loss": 0.0675,
       "step": 8224
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.738844048613044e-06,
-      "loss": 0.0892,
+      "loss": 0.0487,
       "step": 8225
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.736988588922906e-06,
-      "loss": 0.0861,
+      "loss": 0.0416,
       "step": 8226
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.735133129232768e-06,
-      "loss": 0.2989,
+      "loss": 0.0184,
       "step": 8227
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.73327766954263e-06,
-      "loss": 0.1916,
+      "loss": 0.0214,
       "step": 8228
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.731422209852492e-06,
-      "loss": 0.1396,
+      "loss": 0.0612,
       "step": 8229
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.729566750162354e-06,
-      "loss": 0.0875,
+      "loss": 0.051,
       "step": 8230
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.727711290472215e-06,
-      "loss": 0.1372,
+      "loss": 0.0483,
       "step": 8231
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.7258558307820765e-06,
-      "loss": 0.0885,
+      "loss": 0.1374,
       "step": 8232
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.724000371091938e-06,
-      "loss": 0.1404,
+      "loss": 0.1292,
       "step": 8233
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.7221449114018e-06,
-      "loss": 0.086,
+      "loss": 0.1085,
       "step": 8234
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.720289451711662e-06,
-      "loss": 0.1934,
+      "loss": 0.0156,
       "step": 8235
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.718433992021524e-06,
-      "loss": 0.0877,
+      "loss": 0.0138,
       "step": 8236
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.716578532331386e-06,
-      "loss": 0.0854,
+      "loss": 0.0651,
       "step": 8237
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.714723072641248e-06,
-      "loss": 0.1958,
+      "loss": 0.0474,
       "step": 8238
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.712867612951109e-06,
-      "loss": 0.2443,
+      "loss": 0.0193,
       "step": 8239
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.711012153260971e-06,
-      "loss": 0.1953,
+      "loss": 0.126,
       "step": 8240
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.7091566935708325e-06,
-      "loss": 0.1373,
+      "loss": 0.0528,
       "step": 8241
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.707301233880694e-06,
-      "loss": 0.0338,
+      "loss": 0.0579,
       "step": 8242
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.705445774190556e-06,
-      "loss": 0.1936,
+      "loss": 0.1224,
       "step": 8243
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.703590314500418e-06,
-      "loss": 0.0856,
+      "loss": 0.0155,
       "step": 8244
     },
     {
       "epoch": 2.29,
       "learning_rate": 4.70173485481028e-06,
-      "loss": 0.1399,
+      "loss": 0.0432,
       "step": 8245
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.699879395120142e-06,
-      "loss": 0.0851,
+      "loss": 0.0649,
       "step": 8246
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.698023935430004e-06,
-      "loss": 0.0338,
+      "loss": 0.0791,
       "step": 8247
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.696168475739865e-06,
-      "loss": 0.0343,
+      "loss": 0.0484,
       "step": 8248
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.694313016049727e-06,
-      "loss": 0.0876,
+      "loss": 0.1473,
       "step": 8249
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.6924575563595885e-06,
-      "loss": 0.0862,
+      "loss": 0.1319,
       "step": 8250
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.69060209666945e-06,
-      "loss": 0.0886,
+      "loss": 0.0682,
       "step": 8251
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.688746636979312e-06,
-      "loss": 0.1898,
+      "loss": 0.0063,
       "step": 8252
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.686891177289174e-06,
-      "loss": 0.1399,
+      "loss": 0.0581,
       "step": 8253
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.685035717599036e-06,
-      "loss": 0.2461,
+      "loss": 0.0759,
       "step": 8254
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.683180257908898e-06,
-      "loss": 0.1915,
+      "loss": 0.0753,
       "step": 8255
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.681324798218759e-06,
-      "loss": 0.0317,
+      "loss": 0.0251,
       "step": 8256
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.679469338528621e-06,
-      "loss": 0.1913,
+      "loss": 0.0827,
       "step": 8257
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.677613878838483e-06,
-      "loss": 0.032,
+      "loss": 0.0228,
       "step": 8258
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.6757584191483445e-06,
-      "loss": 0.1411,
+      "loss": 0.0559,
       "step": 8259
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.673902959458206e-06,
-      "loss": 0.2383,
+      "loss": 0.1047,
       "step": 8260
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.672047499768068e-06,
-      "loss": 0.0841,
+      "loss": 0.0487,
       "step": 8261
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.67019204007793e-06,
-      "loss": 0.1445,
+      "loss": 0.1314,
       "step": 8262
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.668336580387792e-06,
-      "loss": 0.084,
+      "loss": 0.0473,
       "step": 8263
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.666481120697654e-06,
-      "loss": 0.1418,
+      "loss": 0.1395,
       "step": 8264
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.664625661007515e-06,
-      "loss": 0.1418,
+      "loss": 0.1428,
       "step": 8265
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.662770201317377e-06,
-      "loss": 0.0298,
+      "loss": 0.1422,
       "step": 8266
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.660914741627239e-06,
-      "loss": 0.0872,
+      "loss": 0.0427,
       "step": 8267
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.6590592819371005e-06,
-      "loss": 0.1904,
+      "loss": 0.0902,
       "step": 8268
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.657203822246962e-06,
-      "loss": 0.137,
+      "loss": 0.0948,
       "step": 8269
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.655348362556824e-06,
-      "loss": 0.1395,
+      "loss": 0.118,
       "step": 8270
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.653492902866686e-06,
-      "loss": 0.1391,
+      "loss": 0.0152,
       "step": 8271
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.651637443176548e-06,
-      "loss": 0.086,
+      "loss": 0.1766,
       "step": 8272
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.649781983486409e-06,
-      "loss": 0.2482,
+      "loss": 0.0182,
       "step": 8273
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.647926523796271e-06,
-      "loss": 0.1416,
+      "loss": 0.0465,
       "step": 8274
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.646071064106133e-06,
-      "loss": 0.1906,
+      "loss": 0.0543,
       "step": 8275
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.644215604415994e-06,
-      "loss": 0.0295,
+      "loss": 0.1241,
       "step": 8276
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.642360144725856e-06,
-      "loss": 0.1915,
+      "loss": 0.06,
       "step": 8277
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.6405046850357175e-06,
-      "loss": 0.1981,
+      "loss": 0.1401,
       "step": 8278
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.638649225345579e-06,
-      "loss": 0.1424,
+      "loss": 0.0219,
       "step": 8279
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.636793765655441e-06,
-      "loss": 0.1882,
+      "loss": 0.0956,
       "step": 8280
     },
     {
       "epoch": 2.3,
       "learning_rate": 4.634938305965303e-06,
-      "loss": 0.1935,
+      "loss": 0.0583,
       "step": 8281
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.633082846275165e-06,
-      "loss": 0.0855,
+      "loss": 0.1663,
       "step": 8282
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.631227386585027e-06,
-      "loss": 0.2489,
+      "loss": 0.0964,
       "step": 8283
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.629371926894889e-06,
-      "loss": 0.0826,
+      "loss": 0.0857,
       "step": 8284
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.62751646720475e-06,
-      "loss": 0.1394,
+      "loss": 0.0513,
       "step": 8285
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.625661007514612e-06,
-      "loss": 0.2967,
+      "loss": 0.0522,
       "step": 8286
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.6238055478244735e-06,
-      "loss": 0.1375,
+      "loss": 0.1154,
       "step": 8287
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.621950088134335e-06,
-      "loss": 0.1924,
+      "loss": 0.0296,
       "step": 8288
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.620094628444197e-06,
-      "loss": 0.304,
+      "loss": 0.0815,
       "step": 8289
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.618239168754059e-06,
-      "loss": 0.1419,
+      "loss": 0.0911,
       "step": 8290
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.616383709063921e-06,
-      "loss": 0.0872,
+      "loss": 0.0234,
       "step": 8291
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.614528249373783e-06,
-      "loss": 0.2401,
+      "loss": 0.169,
       "step": 8292
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.612672789683645e-06,
-      "loss": 0.1404,
+      "loss": 0.0632,
       "step": 8293
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.610817329993506e-06,
-      "loss": 0.1416,
+      "loss": 0.0587,
       "step": 8294
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.608961870303368e-06,
-      "loss": 0.3035,
+      "loss": 0.027,
       "step": 8295
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.6071064106132295e-06,
-      "loss": 0.1908,
+      "loss": 0.0989,
       "step": 8296
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.605250950923091e-06,
-      "loss": 0.1406,
+      "loss": 0.1468,
       "step": 8297
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.603395491232953e-06,
-      "loss": 0.0879,
+      "loss": 0.0628,
       "step": 8298
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.601540031542815e-06,
-      "loss": 0.192,
+      "loss": 0.0166,
       "step": 8299
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.599684571852677e-06,
-      "loss": 0.0347,
+      "loss": 0.1272,
       "step": 8300
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.597829112162539e-06,
-      "loss": 0.0869,
+      "loss": 0.0522,
       "step": 8301
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.5959736524724e-06,
-      "loss": 0.0343,
+      "loss": 0.124,
       "step": 8302
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.594118192782262e-06,
-      "loss": 0.0865,
+      "loss": 0.0132,
       "step": 8303
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.592262733092124e-06,
-      "loss": 0.0868,
+      "loss": 0.0248,
       "step": 8304
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.5904072734019855e-06,
-      "loss": 0.1385,
+      "loss": 0.1198,
       "step": 8305
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.588551813711847e-06,
-      "loss": 0.1919,
+      "loss": 0.0666,
       "step": 8306
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.586696354021709e-06,
-      "loss": 0.0887,
+      "loss": 0.0303,
       "step": 8307
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.584840894331571e-06,
-      "loss": 0.0885,
+      "loss": 0.0677,
       "step": 8308
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.582985434641433e-06,
-      "loss": 0.1933,
+      "loss": 0.0589,
       "step": 8309
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.581129974951295e-06,
-      "loss": 0.0335,
+      "loss": 0.054,
       "step": 8310
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.579274515261156e-06,
-      "loss": 0.188,
+      "loss": 0.0628,
       "step": 8311
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.577419055571018e-06,
-      "loss": 0.1431,
+      "loss": 0.0803,
       "step": 8312
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.57556359588088e-06,
-      "loss": 0.0853,
+      "loss": 0.1485,
       "step": 8313
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.5737081361907415e-06,
-      "loss": 0.0332,
+      "loss": 0.0495,
       "step": 8314
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.571852676500603e-06,
-      "loss": 0.1382,
+      "loss": 0.0744,
       "step": 8315
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.569997216810465e-06,
-      "loss": 0.0321,
+      "loss": 0.0127,
       "step": 8316
     },
     {
       "epoch": 2.31,
       "learning_rate": 4.568141757120327e-06,
-      "loss": 0.1421,
+      "loss": 0.1495,
       "step": 8317
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.566286297430189e-06,
-      "loss": 0.1899,
+      "loss": 0.0168,
       "step": 8318
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.56443083774005e-06,
-      "loss": 0.0899,
+      "loss": 0.1149,
       "step": 8319
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.562575378049912e-06,
-      "loss": 0.0328,
+      "loss": 0.1719,
       "step": 8320
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.560719918359774e-06,
-      "loss": 0.0844,
+      "loss": 0.1146,
       "step": 8321
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.5588644586696356e-06,
-      "loss": 0.1348,
+      "loss": 0.1101,
       "step": 8322
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.5570089989794974e-06,
-      "loss": 0.0845,
+      "loss": 0.0151,
       "step": 8323
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.555153539289359e-06,
-      "loss": 0.1924,
+      "loss": 0.1282,
       "step": 8324
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.553298079599221e-06,
-      "loss": 0.0865,
+      "loss": 0.1019,
       "step": 8325
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.551442619909083e-06,
-      "loss": 0.2552,
+      "loss": 0.0307,
       "step": 8326
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.549587160218945e-06,
-      "loss": 0.3077,
+      "loss": 0.0708,
       "step": 8327
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.547731700528806e-06,
-      "loss": 0.0836,
+      "loss": 0.0341,
       "step": 8328
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.545876240838668e-06,
-      "loss": 0.1379,
+      "loss": 0.0674,
       "step": 8329
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.54402078114853e-06,
-      "loss": 0.0869,
+      "loss": 0.0443,
       "step": 8330
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.5421653214583916e-06,
-      "loss": 0.2483,
+      "loss": 0.0829,
       "step": 8331
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.5403098617682534e-06,
-      "loss": 0.1335,
+      "loss": 0.0434,
       "step": 8332
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.538454402078115e-06,
-      "loss": 0.1407,
+      "loss": 0.0155,
       "step": 8333
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.536598942387977e-06,
-      "loss": 0.1918,
+      "loss": 0.119,
       "step": 8334
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.534743482697839e-06,
-      "loss": 0.1407,
+      "loss": 0.0195,
       "step": 8335
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.5328880230077e-06,
-      "loss": 0.1386,
+      "loss": 0.0903,
       "step": 8336
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.531032563317562e-06,
-      "loss": 0.1915,
+      "loss": 0.0195,
       "step": 8337
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.529177103627424e-06,
-      "loss": 0.1406,
+      "loss": 0.0915,
       "step": 8338
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.527321643937286e-06,
-      "loss": 0.0314,
+      "loss": 0.0675,
       "step": 8339
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.5254661842471475e-06,
-      "loss": 0.0833,
+      "loss": 0.1746,
       "step": 8340
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.523610724557009e-06,
-      "loss": 0.3647,
+      "loss": 0.0121,
       "step": 8341
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.521755264866871e-06,
-      "loss": 0.0853,
+      "loss": 0.0657,
       "step": 8342
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.519899805176733e-06,
-      "loss": 0.0845,
+      "loss": 0.017,
       "step": 8343
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.518044345486595e-06,
-      "loss": 0.0318,
+      "loss": 0.108,
       "step": 8344
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.516188885796456e-06,
-      "loss": 0.0867,
+      "loss": 0.0119,
       "step": 8345
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.514333426106318e-06,
-      "loss": 0.0842,
+      "loss": 0.0199,
       "step": 8346
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.51247796641618e-06,
-      "loss": 0.0838,
+      "loss": 0.0876,
       "step": 8347
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.510622506726042e-06,
-      "loss": 0.1399,
+      "loss": 0.0782,
       "step": 8348
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.5087670470359035e-06,
-      "loss": 0.0828,
+      "loss": 0.0824,
       "step": 8349
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.506911587345765e-06,
-      "loss": 0.2521,
+      "loss": 0.0858,
       "step": 8350
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.505056127655627e-06,
-      "loss": 0.1365,
+      "loss": 0.1009,
       "step": 8351
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.503200667965489e-06,
-      "loss": 0.0854,
+      "loss": 0.0561,
       "step": 8352
     },
     {
       "epoch": 2.32,
       "learning_rate": 4.501345208275351e-06,
-      "loss": 0.3061,
+      "loss": 0.0191,
       "step": 8353
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.499489748585212e-06,
-      "loss": 0.0305,
+      "loss": 0.0156,
       "step": 8354
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.497634288895074e-06,
-      "loss": 0.1385,
+      "loss": 0.1611,
       "step": 8355
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.495778829204936e-06,
-      "loss": 0.2458,
+      "loss": 0.1973,
       "step": 8356
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.493923369514798e-06,
-      "loss": 0.0835,
+      "loss": 0.1564,
       "step": 8357
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.4920679098246595e-06,
-      "loss": 0.136,
+      "loss": 0.1578,
       "step": 8358
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.490212450134521e-06,
-      "loss": 0.1384,
+      "loss": 0.0558,
       "step": 8359
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.488356990444383e-06,
-      "loss": 0.0302,
+      "loss": 0.0595,
       "step": 8360
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.486501530754245e-06,
-      "loss": 0.0304,
+      "loss": 0.071,
       "step": 8361
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.484646071064106e-06,
-      "loss": 0.031,
+      "loss": 0.0634,
       "step": 8362
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.482790611373968e-06,
-      "loss": 0.031,
+      "loss": 0.0897,
       "step": 8363
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.48093515168383e-06,
-      "loss": 0.1373,
+      "loss": 0.0143,
       "step": 8364
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.479079691993692e-06,
-      "loss": 0.0289,
+      "loss": 0.0781,
       "step": 8365
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.477224232303554e-06,
-      "loss": 0.084,
+      "loss": 0.0626,
       "step": 8366
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.4753687726134155e-06,
-      "loss": 0.195,
+      "loss": 0.106,
       "step": 8367
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.473513312923277e-06,
-      "loss": 0.0816,
+      "loss": 0.2554,
       "step": 8368
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.471657853233139e-06,
-      "loss": 0.135,
+      "loss": 0.044,
       "step": 8369
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.469802393543001e-06,
-      "loss": 0.1407,
+      "loss": 0.1661,
       "step": 8370
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.467946933852862e-06,
-      "loss": 0.138,
+      "loss": 0.1169,
       "step": 8371
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.466091474162724e-06,
-      "loss": 0.2443,
+      "loss": 0.1235,
       "step": 8372
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.464236014472586e-06,
-      "loss": 0.0283,
+      "loss": 0.0185,
       "step": 8373
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.462380554782448e-06,
-      "loss": 0.0819,
+      "loss": 0.0558,
       "step": 8374
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.46052509509231e-06,
-      "loss": 0.2472,
+      "loss": 0.0523,
       "step": 8375
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.4586696354021715e-06,
-      "loss": 0.0863,
+      "loss": 0.1429,
       "step": 8376
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.456814175712033e-06,
-      "loss": 0.0286,
+      "loss": 0.0539,
       "step": 8377
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.454958716021895e-06,
-      "loss": 0.1954,
+      "loss": 0.0324,
       "step": 8378
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.453103256331756e-06,
-      "loss": 0.0277,
+      "loss": 0.0271,
       "step": 8379
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.451247796641618e-06,
-      "loss": 0.2509,
+      "loss": 0.0795,
       "step": 8380
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.44939233695148e-06,
-      "loss": 0.1932,
+      "loss": 0.0269,
       "step": 8381
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.447536877261342e-06,
-      "loss": 0.0846,
+      "loss": 0.0606,
       "step": 8382
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.445681417571204e-06,
-      "loss": 0.1946,
+      "loss": 0.0713,
       "step": 8383
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.443825957881066e-06,
-      "loss": 0.1931,
+      "loss": 0.0586,
       "step": 8384
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.4419704981909275e-06,
-      "loss": 0.0846,
+      "loss": 0.0501,
       "step": 8385
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.440115038500789e-06,
-      "loss": 0.1956,
+      "loss": 0.0181,
       "step": 8386
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.438259578810651e-06,
-      "loss": 0.1917,
+      "loss": 0.0784,
       "step": 8387
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.436404119120512e-06,
-      "loss": 0.1925,
+      "loss": 0.0458,
       "step": 8388
     },
     {
       "epoch": 2.33,
       "learning_rate": 4.434548659430374e-06,
-      "loss": 0.0291,
+      "loss": 0.0114,
       "step": 8389
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.432693199740236e-06,
-      "loss": 0.1415,
+      "loss": 0.0172,
       "step": 8390
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.430837740050098e-06,
-      "loss": 0.1409,
+      "loss": 0.0151,
       "step": 8391
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.42898228035996e-06,
-      "loss": 0.0806,
+      "loss": 0.0114,
       "step": 8392
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.427126820669822e-06,
-      "loss": 0.0823,
+      "loss": 0.1428,
       "step": 8393
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.4252713609796835e-06,
-      "loss": 0.1918,
+      "loss": 0.2414,
       "step": 8394
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.423415901289545e-06,
-      "loss": 0.1386,
+      "loss": 0.1461,
       "step": 8395
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.421560441599406e-06,
-      "loss": 0.1954,
+      "loss": 0.0136,
       "step": 8396
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.419704981909268e-06,
-      "loss": 0.1387,
+      "loss": 0.0059,
       "step": 8397
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.41784952221913e-06,
-      "loss": 0.1964,
+      "loss": 0.0162,
       "step": 8398
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.415994062528992e-06,
-      "loss": 0.2511,
+      "loss": 0.0191,
       "step": 8399
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.414138602838854e-06,
-      "loss": 0.0842,
+      "loss": 0.1202,
       "step": 8400
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.412283143148716e-06,
-      "loss": 0.1395,
+      "loss": 0.0085,
       "step": 8401
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.410427683458578e-06,
-      "loss": 0.249,
+      "loss": 0.0545,
       "step": 8402
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.4085722237684395e-06,
-      "loss": 0.1387,
+      "loss": 0.0677,
       "step": 8403
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.406716764078301e-06,
-      "loss": 0.1403,
+      "loss": 0.0181,
       "step": 8404
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.404861304388162e-06,
-      "loss": 0.296,
+      "loss": 0.0133,
       "step": 8405
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.403005844698024e-06,
-      "loss": 0.0297,
+      "loss": 0.0214,
       "step": 8406
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.401150385007886e-06,
-      "loss": 0.2976,
+      "loss": 0.1637,
       "step": 8407
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.399294925317748e-06,
-      "loss": 0.0834,
+      "loss": 0.0123,
       "step": 8408
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.39743946562761e-06,
-      "loss": 0.1918,
+      "loss": 0.052,
       "step": 8409
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.395584005937472e-06,
-      "loss": 0.0849,
+      "loss": 0.0533,
       "step": 8410
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.393728546247334e-06,
-      "loss": 0.1936,
+      "loss": 0.0217,
       "step": 8411
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.3918730865571955e-06,
-      "loss": 0.1964,
+      "loss": 0.0193,
       "step": 8412
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.390017626867057e-06,
-      "loss": 0.1411,
+      "loss": 0.0748,
       "step": 8413
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.388162167176918e-06,
-      "loss": 0.1388,
+      "loss": 0.0093,
       "step": 8414
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.38630670748678e-06,
-      "loss": 0.0321,
+      "loss": 0.0619,
       "step": 8415
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.384451247796642e-06,
-      "loss": 0.1405,
+      "loss": 0.0155,
       "step": 8416
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.382595788106504e-06,
-      "loss": 0.14,
+      "loss": 0.0656,
       "step": 8417
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.380740328416366e-06,
-      "loss": 0.0843,
+      "loss": 0.0056,
       "step": 8418
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.378884868726228e-06,
-      "loss": 0.1929,
+      "loss": 0.1353,
       "step": 8419
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.3770294090360896e-06,
-      "loss": 0.0328,
+      "loss": 0.046,
       "step": 8420
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.3751739493459514e-06,
-      "loss": 0.0321,
+      "loss": 0.0416,
       "step": 8421
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.3733184896558125e-06,
-      "loss": 0.0846,
+      "loss": 0.0154,
       "step": 8422
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.371463029965674e-06,
-      "loss": 0.141,
+      "loss": 0.0569,
       "step": 8423
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.369607570275536e-06,
-      "loss": 0.1394,
+      "loss": 0.0733,
       "step": 8424
     },
     {
       "epoch": 2.34,
       "learning_rate": 4.367752110585398e-06,
-      "loss": 0.138,
+      "loss": 0.1984,
       "step": 8425
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.365896650895259e-06,
-      "loss": 0.1389,
+      "loss": 0.0166,
       "step": 8426
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.364041191205121e-06,
-      "loss": 0.0321,
+      "loss": 0.138,
       "step": 8427
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.362185731514983e-06,
-      "loss": 0.1438,
+      "loss": 0.0096,
       "step": 8428
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.360330271824845e-06,
-      "loss": 0.0858,
+      "loss": 0.0508,
       "step": 8429
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.358474812134707e-06,
-      "loss": 0.1929,
+      "loss": 0.1432,
       "step": 8430
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.3566193524445685e-06,
-      "loss": 0.0313,
+      "loss": 0.0683,
       "step": 8431
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.35476389275443e-06,
-      "loss": 0.083,
+      "loss": 0.0931,
       "step": 8432
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.352908433064292e-06,
-      "loss": 0.0316,
+      "loss": 0.1203,
       "step": 8433
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.351052973374153e-06,
-      "loss": 0.1388,
+      "loss": 0.0691,
       "step": 8434
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.349197513684015e-06,
-      "loss": 0.1404,
+      "loss": 0.0118,
       "step": 8435
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.347342053993877e-06,
-      "loss": 0.2531,
+      "loss": 0.0905,
       "step": 8436
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.345486594303739e-06,
-      "loss": 0.0298,
+      "loss": 0.1581,
       "step": 8437
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.343631134613601e-06,
-      "loss": 0.0834,
+      "loss": 0.023,
       "step": 8438
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.3417756749234626e-06,
-      "loss": 0.0843,
+      "loss": 0.0161,
       "step": 8439
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.3399202152333244e-06,
-      "loss": 0.2516,
+      "loss": 0.0114,
       "step": 8440
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.338064755543186e-06,
-      "loss": 0.0811,
+      "loss": 0.0425,
       "step": 8441
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.336209295853047e-06,
-      "loss": 0.1941,
+      "loss": 0.106,
       "step": 8442
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.334353836162909e-06,
-      "loss": 0.1417,
+      "loss": 0.0366,
       "step": 8443
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.332498376472771e-06,
-      "loss": 0.0834,
+      "loss": 0.0663,
       "step": 8444
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.330642916782633e-06,
-      "loss": 0.1413,
+      "loss": 0.1052,
       "step": 8445
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.328787457092495e-06,
-      "loss": 0.0857,
+      "loss": 0.1442,
       "step": 8446
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.326931997402357e-06,
-      "loss": 0.0299,
+      "loss": 0.0169,
       "step": 8447
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.3250765377122186e-06,
-      "loss": 0.082,
+      "loss": 0.0792,
       "step": 8448
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.3232210780220804e-06,
-      "loss": 0.1416,
+      "loss": 0.045,
       "step": 8449
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.321365618331942e-06,
-      "loss": 0.0291,
+      "loss": 0.1144,
       "step": 8450
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.319510158641803e-06,
-      "loss": 0.2501,
+      "loss": 0.0601,
       "step": 8451
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.317654698951665e-06,
-      "loss": 0.0291,
+      "loss": 0.0479,
       "step": 8452
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.315799239261527e-06,
-      "loss": 0.2486,
+      "loss": 0.0671,
       "step": 8453
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.313943779571389e-06,
-      "loss": 0.2525,
+      "loss": 0.0539,
       "step": 8454
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.312088319881251e-06,
-      "loss": 0.1407,
+      "loss": 0.052,
       "step": 8455
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.310232860191113e-06,
-      "loss": 0.1392,
+      "loss": 0.1594,
       "step": 8456
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.3083774005009745e-06,
-      "loss": 0.0838,
+      "loss": 0.0237,
       "step": 8457
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.306521940810836e-06,
-      "loss": 0.1954,
+      "loss": 0.0658,
       "step": 8458
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.3046664811206974e-06,
-      "loss": 0.1945,
+      "loss": 0.1024,
       "step": 8459
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.302811021430559e-06,
-      "loss": 0.1397,
+      "loss": 0.0131,
       "step": 8460
     },
     {
       "epoch": 2.35,
       "learning_rate": 4.300955561740421e-06,
-      "loss": 0.1391,
+      "loss": 0.0649,
       "step": 8461
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.299100102050283e-06,
-      "loss": 0.0859,
+      "loss": 0.0118,
       "step": 8462
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.297244642360145e-06,
-      "loss": 0.1938,
+      "loss": 0.1915,
       "step": 8463
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.295389182670007e-06,
-      "loss": 0.0882,
+      "loss": 0.0187,
       "step": 8464
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.293533722979869e-06,
-      "loss": 0.14,
+      "loss": 0.1693,
       "step": 8465
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.2916782632897305e-06,
-      "loss": 0.084,
+      "loss": 0.1352,
       "step": 8466
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.289822803599592e-06,
-      "loss": 0.0294,
+      "loss": 0.0145,
       "step": 8467
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.2879673439094534e-06,
-      "loss": 0.1951,
+      "loss": 0.2196,
       "step": 8468
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.286111884219315e-06,
-      "loss": 0.0296,
+      "loss": 0.0238,
       "step": 8469
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.284256424529177e-06,
-      "loss": 0.3058,
+      "loss": 0.0536,
       "step": 8470
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.282400964839039e-06,
-      "loss": 0.029,
+      "loss": 0.061,
       "step": 8471
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.280545505148901e-06,
-      "loss": 0.0851,
+      "loss": 0.0837,
       "step": 8472
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.278690045458763e-06,
-      "loss": 0.2513,
+      "loss": 0.0222,
       "step": 8473
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.276834585768625e-06,
-      "loss": 0.0857,
+      "loss": 0.0223,
       "step": 8474
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.2749791260784865e-06,
-      "loss": 0.0281,
+      "loss": 0.0302,
       "step": 8475
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.2731236663883475e-06,
-      "loss": 0.1355,
+      "loss": 0.0241,
       "step": 8476
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.271268206698209e-06,
-      "loss": 0.195,
+      "loss": 0.0175,
       "step": 8477
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.269412747008071e-06,
-      "loss": 0.3571,
+      "loss": 0.0607,
       "step": 8478
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.267557287317933e-06,
-      "loss": 0.1951,
+      "loss": 0.0419,
       "step": 8479
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.265701827627795e-06,
-      "loss": 0.1424,
+      "loss": 0.1054,
       "step": 8480
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.263846367937657e-06,
-      "loss": 0.1367,
+      "loss": 0.0618,
       "step": 8481
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.261990908247519e-06,
-      "loss": 0.0842,
+      "loss": 0.0629,
       "step": 8482
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.260135448557381e-06,
-      "loss": 0.0292,
+      "loss": 0.0488,
       "step": 8483
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.2582799888672425e-06,
-      "loss": 0.1416,
+      "loss": 0.2612,
       "step": 8484
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.2564245291771035e-06,
-      "loss": 0.0849,
+      "loss": 0.0894,
       "step": 8485
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.254569069486965e-06,
-      "loss": 0.1923,
+      "loss": 0.0469,
       "step": 8486
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.252713609796827e-06,
-      "loss": 0.1382,
+      "loss": 0.1228,
       "step": 8487
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.250858150106689e-06,
-      "loss": 0.0294,
+      "loss": 0.0786,
       "step": 8488
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.249002690416551e-06,
-      "loss": 0.0845,
+      "loss": 0.0263,
       "step": 8489
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.247147230726413e-06,
-      "loss": 0.0843,
+      "loss": 0.0152,
       "step": 8490
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.245291771036275e-06,
-      "loss": 0.1399,
+      "loss": 0.0815,
       "step": 8491
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.243436311346137e-06,
-      "loss": 0.0291,
+      "loss": 0.1107,
       "step": 8492
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.2415808516559985e-06,
-      "loss": 0.0297,
+      "loss": 0.054,
       "step": 8493
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.2397253919658595e-06,
-      "loss": 0.1368,
+      "loss": 0.0584,
       "step": 8494
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.237869932275721e-06,
-      "loss": 0.1976,
+      "loss": 0.1012,
       "step": 8495
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.236014472585583e-06,
-      "loss": 0.1421,
+      "loss": 0.1014,
       "step": 8496
     },
     {
       "epoch": 2.36,
       "learning_rate": 4.234159012895445e-06,
-      "loss": 0.1952,
+      "loss": 0.0689,
       "step": 8497
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.232303553205307e-06,
-      "loss": 0.0832,
+      "loss": 0.0547,
       "step": 8498
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.230448093515169e-06,
-      "loss": 0.0814,
+      "loss": 0.0979,
       "step": 8499
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.228592633825031e-06,
-      "loss": 0.1968,
+      "loss": 0.2425,
       "step": 8500
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.226737174134893e-06,
-      "loss": 0.0292,
+      "loss": 0.067,
       "step": 8501
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.224881714444754e-06,
-      "loss": 0.0808,
+      "loss": 0.1003,
       "step": 8502
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.2230262547546155e-06,
-      "loss": 0.1381,
+      "loss": 0.0707,
       "step": 8503
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.221170795064477e-06,
-      "loss": 0.1936,
+      "loss": 0.0596,
       "step": 8504
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.219315335374339e-06,
-      "loss": 0.0272,
+      "loss": 0.0453,
       "step": 8505
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.217459875684201e-06,
-      "loss": 0.1428,
+      "loss": 0.018,
       "step": 8506
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.215604415994063e-06,
-      "loss": 0.2512,
+      "loss": 0.1257,
       "step": 8507
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.213748956303925e-06,
-      "loss": 0.1958,
+      "loss": 0.0876,
       "step": 8508
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.211893496613787e-06,
-      "loss": 0.136,
+      "loss": 0.0224,
       "step": 8509
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.210038036923649e-06,
-      "loss": 0.2417,
+      "loss": 0.1022,
       "step": 8510
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.20818257723351e-06,
-      "loss": 0.0843,
+      "loss": 0.0758,
       "step": 8511
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.2063271175433715e-06,
-      "loss": 0.202,
+      "loss": 0.0123,
       "step": 8512
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.204471657853233e-06,
-      "loss": 0.1415,
+      "loss": 0.0892,
       "step": 8513
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.202616198163095e-06,
-      "loss": 0.138,
+      "loss": 0.0477,
       "step": 8514
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.200760738472957e-06,
-      "loss": 0.0829,
+      "loss": 0.0899,
       "step": 8515
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.198905278782819e-06,
-      "loss": 0.2468,
+      "loss": 0.027,
       "step": 8516
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.197049819092681e-06,
-      "loss": 0.0835,
+      "loss": 0.0179,
       "step": 8517
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.195194359402543e-06,
-      "loss": 0.3052,
+      "loss": 0.0523,
       "step": 8518
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.193338899712404e-06,
-      "loss": 0.085,
+      "loss": 0.0832,
       "step": 8519
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.191483440022266e-06,
-      "loss": 0.137,
+      "loss": 0.1262,
       "step": 8520
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.1896279803321275e-06,
-      "loss": 0.0861,
+      "loss": 0.0952,
       "step": 8521
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.187772520641989e-06,
-      "loss": 0.0866,
+      "loss": 0.0173,
       "step": 8522
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.185917060951851e-06,
-      "loss": 0.1389,
+      "loss": 0.0154,
       "step": 8523
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.184061601261713e-06,
-      "loss": 0.1948,
+      "loss": 0.044,
       "step": 8524
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.182206141571575e-06,
-      "loss": 0.086,
+      "loss": 0.041,
       "step": 8525
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.180350681881437e-06,
-      "loss": 0.1938,
+      "loss": 0.0434,
       "step": 8526
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.178495222191299e-06,
-      "loss": 0.086,
+      "loss": 0.113,
       "step": 8527
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.17663976250116e-06,
-      "loss": 0.0298,
+      "loss": 0.1646,
       "step": 8528
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.174784302811022e-06,
-      "loss": 0.0304,
+      "loss": 0.0924,
       "step": 8529
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.1729288431208835e-06,
-      "loss": 0.1393,
+      "loss": 0.0202,
       "step": 8530
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.171073383430745e-06,
-      "loss": 0.0836,
+      "loss": 0.0083,
       "step": 8531
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.169217923740607e-06,
-      "loss": 0.133,
+      "loss": 0.1156,
       "step": 8532
     },
     {
       "epoch": 2.37,
       "learning_rate": 4.167362464050469e-06,
-      "loss": 0.0853,
+      "loss": 0.0796,
       "step": 8533
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.165507004360331e-06,
-      "loss": 0.03,
+      "loss": 0.0452,
       "step": 8534
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.163651544670193e-06,
-      "loss": 0.2492,
+      "loss": 0.0473,
       "step": 8535
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.161796084980054e-06,
-      "loss": 0.0849,
+      "loss": 0.0398,
       "step": 8536
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.159940625289916e-06,
-      "loss": 0.1387,
+      "loss": 0.0873,
       "step": 8537
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.158085165599778e-06,
-      "loss": 0.25,
+      "loss": 0.018,
       "step": 8538
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.1562297059096395e-06,
-      "loss": 0.0842,
+      "loss": 0.1461,
       "step": 8539
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.154374246219501e-06,
-      "loss": 0.138,
+      "loss": 0.1114,
       "step": 8540
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.152518786529363e-06,
-      "loss": 0.138,
+      "loss": 0.1568,
       "step": 8541
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.150663326839225e-06,
-      "loss": 0.2475,
+      "loss": 0.0579,
       "step": 8542
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.148807867149087e-06,
-      "loss": 0.3132,
+      "loss": 0.1634,
       "step": 8543
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.146952407458949e-06,
-      "loss": 0.1416,
+      "loss": 0.1149,
       "step": 8544
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.14509694776881e-06,
-      "loss": 0.1393,
+      "loss": 0.0193,
       "step": 8545
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.143241488078672e-06,
-      "loss": 0.3058,
+      "loss": 0.0175,
       "step": 8546
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.141386028388534e-06,
-      "loss": 0.1397,
+      "loss": 0.0411,
       "step": 8547
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.1395305686983955e-06,
-      "loss": 0.1359,
+      "loss": 0.0472,
       "step": 8548
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.137675109008257e-06,
-      "loss": 0.0859,
+      "loss": 0.0127,
       "step": 8549
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.135819649318119e-06,
-      "loss": 0.1391,
+      "loss": 0.0484,
       "step": 8550
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.133964189627981e-06,
-      "loss": 0.141,
+      "loss": 0.0786,
       "step": 8551
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.132108729937843e-06,
-      "loss": 0.1388,
+      "loss": 0.0283,
       "step": 8552
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.130253270247705e-06,
-      "loss": 0.0304,
+      "loss": 0.1319,
       "step": 8553
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.128397810557566e-06,
-      "loss": 0.0848,
+      "loss": 0.0461,
       "step": 8554
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.126542350867428e-06,
-      "loss": 0.138,
+      "loss": 0.0751,
       "step": 8555
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.1246868911772896e-06,
-      "loss": 0.1974,
+      "loss": 0.1104,
       "step": 8556
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.1228314314871514e-06,
-      "loss": 0.0308,
+      "loss": 0.0381,
       "step": 8557
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.120975971797013e-06,
-      "loss": 0.135,
+      "loss": 0.0258,
       "step": 8558
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.119120512106875e-06,
-      "loss": 0.1961,
+      "loss": 0.0172,
       "step": 8559
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.117265052416737e-06,
-      "loss": 0.0839,
+      "loss": 0.0994,
       "step": 8560
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.115409592726599e-06,
-      "loss": 0.0823,
+      "loss": 0.197,
       "step": 8561
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.11355413303646e-06,
-      "loss": 0.0858,
+      "loss": 0.0834,
       "step": 8562
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.111698673346322e-06,
-      "loss": 0.194,
+      "loss": 0.1111,
       "step": 8563
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.109843213656184e-06,
-      "loss": 0.1929,
+      "loss": 0.1588,
       "step": 8564
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.1079877539660456e-06,
-      "loss": 0.0298,
+      "loss": 0.1051,
       "step": 8565
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.1061322942759074e-06,
-      "loss": 0.086,
+      "loss": 0.0729,
       "step": 8566
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.104276834585769e-06,
-      "loss": 0.0856,
+      "loss": 0.0784,
       "step": 8567
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.102421374895631e-06,
-      "loss": 0.194,
+      "loss": 0.1464,
       "step": 8568
     },
     {
       "epoch": 2.38,
       "learning_rate": 4.100565915205493e-06,
-      "loss": 0.0307,
+      "loss": 0.326,
       "step": 8569
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.098710455515355e-06,
-      "loss": 0.1406,
+      "loss": 0.02,
       "step": 8570
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.096854995825216e-06,
-      "loss": 0.1928,
+      "loss": 0.1019,
       "step": 8571
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.094999536135078e-06,
-      "loss": 0.1391,
+      "loss": 0.0527,
       "step": 8572
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.09314407644494e-06,
-      "loss": 0.0862,
+      "loss": 0.0741,
       "step": 8573
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.0912886167548015e-06,
-      "loss": 0.0294,
+      "loss": 0.0254,
       "step": 8574
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.089433157064663e-06,
-      "loss": 0.1906,
+      "loss": 0.0233,
       "step": 8575
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.087577697374525e-06,
-      "loss": 0.2511,
+      "loss": 0.0185,
       "step": 8576
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.085722237684387e-06,
-      "loss": 0.137,
+      "loss": 0.0156,
       "step": 8577
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.083866777994248e-06,
-      "loss": 0.3507,
+      "loss": 0.0964,
       "step": 8578
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.08201131830411e-06,
-      "loss": 0.0867,
+      "loss": 0.0266,
       "step": 8579
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.080155858613972e-06,
-      "loss": 0.0303,
+      "loss": 0.0598,
       "step": 8580
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.078300398923834e-06,
-      "loss": 0.1356,
+      "loss": 0.0656,
       "step": 8581
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.076444939233695e-06,
-      "loss": 0.142,
+      "loss": 0.0939,
       "step": 8582
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.074589479543557e-06,
-      "loss": 0.1914,
+      "loss": 0.0985,
       "step": 8583
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.0727340198534186e-06,
-      "loss": 0.1404,
+      "loss": 0.0771,
       "step": 8584
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.0708785601632804e-06,
-      "loss": 0.0875,
+      "loss": 0.0747,
       "step": 8585
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.069023100473142e-06,
-      "loss": 0.1415,
+      "loss": 0.1104,
       "step": 8586
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.067167640783004e-06,
-      "loss": 0.087,
+      "loss": 0.1063,
       "step": 8587
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.065312181092866e-06,
-      "loss": 0.19,
+      "loss": 0.0181,
       "step": 8588
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.063456721402728e-06,
-      "loss": 0.193,
+      "loss": 0.013,
       "step": 8589
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.06160126171259e-06,
-      "loss": 0.0866,
+      "loss": 0.0221,
       "step": 8590
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.059745802022451e-06,
-      "loss": 0.2524,
+      "loss": 0.151,
       "step": 8591
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.057890342332313e-06,
-      "loss": 0.1928,
+      "loss": 0.2587,
       "step": 8592
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.0560348826421745e-06,
-      "loss": 0.0855,
+      "loss": 0.0198,
       "step": 8593
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.054179422952036e-06,
-      "loss": 0.1411,
+      "loss": 0.0611,
       "step": 8594
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.052323963261898e-06,
-      "loss": 0.1425,
+      "loss": 0.0606,
       "step": 8595
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.05046850357176e-06,
-      "loss": 0.2434,
+      "loss": 0.0193,
       "step": 8596
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.048613043881622e-06,
-      "loss": 0.0855,
+      "loss": 0.0345,
       "step": 8597
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.046757584191484e-06,
-      "loss": 0.139,
+      "loss": 0.0649,
       "step": 8598
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.044902124501345e-06,
-      "loss": 0.1362,
+      "loss": 0.0107,
       "step": 8599
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.043046664811207e-06,
-      "loss": 0.1975,
+      "loss": 0.1149,
       "step": 8600
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.041191205121069e-06,
-      "loss": 0.1937,
+      "loss": 0.0212,
       "step": 8601
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.0393357454309305e-06,
-      "loss": 0.192,
+      "loss": 0.0139,
       "step": 8602
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.037480285740792e-06,
-      "loss": 0.0843,
+      "loss": 0.0457,
       "step": 8603
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.035624826050654e-06,
-      "loss": 0.139,
+      "loss": 0.074,
       "step": 8604
     },
     {
       "epoch": 2.39,
       "learning_rate": 4.033769366360516e-06,
-      "loss": 0.1391,
+      "loss": 0.1041,
       "step": 8605
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.031913906670378e-06,
-      "loss": 0.1425,
+      "loss": 0.055,
       "step": 8606
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.03005844698024e-06,
-      "loss": 0.1405,
+      "loss": 0.0583,
       "step": 8607
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.028202987290101e-06,
-      "loss": 0.2967,
+      "loss": 0.0553,
       "step": 8608
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.026347527599963e-06,
-      "loss": 0.0316,
+      "loss": 0.0142,
       "step": 8609
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.024492067909825e-06,
-      "loss": 0.0857,
+      "loss": 0.0934,
       "step": 8610
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.0226366082196865e-06,
-      "loss": 0.1408,
+      "loss": 0.0563,
       "step": 8611
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.020781148529548e-06,
-      "loss": 0.19,
+      "loss": 0.0272,
       "step": 8612
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.01892568883941e-06,
-      "loss": 0.0847,
+      "loss": 0.1089,
       "step": 8613
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.017070229149272e-06,
-      "loss": 0.1935,
+      "loss": 0.0803,
       "step": 8614
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.015214769459134e-06,
-      "loss": 0.0865,
+      "loss": 0.0195,
       "step": 8615
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.013359309768996e-06,
-      "loss": 0.2429,
+      "loss": 0.0246,
       "step": 8616
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.011503850078857e-06,
-      "loss": 0.1909,
+      "loss": 0.0654,
       "step": 8617
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.009648390388719e-06,
-      "loss": 0.1399,
+      "loss": 0.086,
       "step": 8618
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.007792930698581e-06,
-      "loss": 0.1921,
+      "loss": 0.0541,
       "step": 8619
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.0059374710084425e-06,
-      "loss": 0.0322,
+      "loss": 0.0438,
       "step": 8620
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.004082011318304e-06,
-      "loss": 0.1368,
+      "loss": 0.0571,
       "step": 8621
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.002226551628166e-06,
-      "loss": 0.1901,
+      "loss": 0.0414,
       "step": 8622
     },
     {
       "epoch": 2.4,
       "learning_rate": 4.000371091938028e-06,
-      "loss": 0.2873,
+      "loss": 0.071,
       "step": 8623
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.99851563224789e-06,
-      "loss": 0.1932,
+      "loss": 0.1629,
       "step": 8624
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.996660172557751e-06,
-      "loss": 0.3024,
+      "loss": 0.015,
       "step": 8625
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.994804712867613e-06,
-      "loss": 0.1393,
+      "loss": 0.0415,
       "step": 8626
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.992949253177475e-06,
-      "loss": 0.0875,
+      "loss": 0.0949,
       "step": 8627
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.991093793487337e-06,
-      "loss": 0.1953,
+      "loss": 0.0908,
       "step": 8628
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.9892383337971985e-06,
-      "loss": 0.0874,
+      "loss": 0.0447,
       "step": 8629
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.98738287410706e-06,
-      "loss": 0.1364,
+      "loss": 0.0061,
       "step": 8630
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.985527414416922e-06,
-      "loss": 0.2942,
+      "loss": 0.0786,
       "step": 8631
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.983671954726784e-06,
-      "loss": 0.0339,
+      "loss": 0.14,
       "step": 8632
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.981816495036646e-06,
-      "loss": 0.1382,
+      "loss": 0.0439,
       "step": 8633
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.979961035346507e-06,
-      "loss": 0.088,
+      "loss": 0.0101,
       "step": 8634
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.978105575656369e-06,
-      "loss": 0.2419,
+      "loss": 0.0182,
       "step": 8635
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.976250115966231e-06,
-      "loss": 0.0364,
+      "loss": 0.0496,
       "step": 8636
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.974394656276093e-06,
-      "loss": 0.0833,
+      "loss": 0.0222,
       "step": 8637
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.9725391965859545e-06,
-      "loss": 0.0855,
+      "loss": 0.0217,
       "step": 8638
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.970683736895816e-06,
-      "loss": 0.0361,
+      "loss": 0.0315,
       "step": 8639
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.968828277205678e-06,
-      "loss": 0.1931,
+      "loss": 0.0815,
       "step": 8640
     },
     {
       "epoch": 2.4,
       "learning_rate": 3.96697281751554e-06,
-      "loss": 0.0876,
+      "loss": 0.0791,
       "step": 8641
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.965117357825401e-06,
-      "loss": 0.2462,
+      "loss": 0.0077,
       "step": 8642
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.963261898135263e-06,
-      "loss": 0.1904,
+      "loss": 0.1225,
       "step": 8643
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.961406438445125e-06,
-      "loss": 0.1926,
+      "loss": 0.0345,
       "step": 8644
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.959550978754987e-06,
-      "loss": 0.1374,
+      "loss": 0.0332,
       "step": 8645
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.957695519064849e-06,
-      "loss": 0.1921,
+      "loss": 0.0875,
       "step": 8646
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.9558400593747105e-06,
-      "loss": 0.0327,
+      "loss": 0.1431,
       "step": 8647
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.953984599684572e-06,
-      "loss": 0.139,
+      "loss": 0.0863,
       "step": 8648
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.952129139994434e-06,
-      "loss": 0.0832,
+      "loss": 0.0461,
       "step": 8649
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.950273680304296e-06,
-      "loss": 0.1383,
+      "loss": 0.2593,
       "step": 8650
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.948418220614157e-06,
-      "loss": 0.0869,
+      "loss": 0.1535,
       "step": 8651
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.946562760924019e-06,
-      "loss": 0.1923,
+      "loss": 0.1302,
       "step": 8652
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.944707301233881e-06,
-      "loss": 0.192,
+      "loss": 0.0395,
       "step": 8653
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.942851841543743e-06,
-      "loss": 0.0863,
+      "loss": 0.1686,
       "step": 8654
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.940996381853605e-06,
-      "loss": 0.4056,
+      "loss": 0.0457,
       "step": 8655
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.9391409221634665e-06,
-      "loss": 0.1927,
+      "loss": 0.2127,
       "step": 8656
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.937285462473328e-06,
-      "loss": 0.1893,
+      "loss": 0.1328,
       "step": 8657
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.93543000278319e-06,
-      "loss": 0.1393,
+      "loss": 0.0862,
       "step": 8658
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.933574543093051e-06,
-      "loss": 0.1368,
+      "loss": 0.111,
       "step": 8659
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.931719083402913e-06,
-      "loss": 0.1393,
+      "loss": 0.1001,
       "step": 8660
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.929863623712775e-06,
-      "loss": 0.1382,
+      "loss": 0.0115,
       "step": 8661
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.928008164022637e-06,
-      "loss": 0.1883,
+      "loss": 0.0121,
       "step": 8662
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.926152704332499e-06,
-      "loss": 0.0829,
+      "loss": 0.0573,
       "step": 8663
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.924297244642361e-06,
-      "loss": 0.1382,
+      "loss": 0.1151,
       "step": 8664
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.9224417849522225e-06,
-      "loss": 0.1399,
+      "loss": 0.0592,
       "step": 8665
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.920586325262084e-06,
-      "loss": 0.0869,
+      "loss": 0.0552,
       "step": 8666
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.918730865571946e-06,
-      "loss": 0.0347,
+      "loss": 0.0301,
       "step": 8667
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.916875405881807e-06,
-      "loss": 0.0346,
+      "loss": 0.0673,
       "step": 8668
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.915019946191669e-06,
-      "loss": 0.1913,
+      "loss": 0.0565,
       "step": 8669
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.913164486501531e-06,
-      "loss": 0.2448,
+      "loss": 0.1266,
       "step": 8670
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.911309026811393e-06,
-      "loss": 0.0881,
+      "loss": 0.0469,
       "step": 8671
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.909453567121255e-06,
-      "loss": 0.1905,
+      "loss": 0.08,
       "step": 8672
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.9075981074311166e-06,
-      "loss": 0.0332,
+      "loss": 0.0806,
       "step": 8673
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.9057426477409784e-06,
-      "loss": 0.087,
+      "loss": 0.0187,
       "step": 8674
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.90388718805084e-06,
-      "loss": 0.296,
+      "loss": 0.0679,
       "step": 8675
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.902031728360702e-06,
-      "loss": 0.1907,
+      "loss": 0.0396,
       "step": 8676
     },
     {
       "epoch": 2.41,
       "learning_rate": 3.900176268670563e-06,
-      "loss": 0.1364,
+      "loss": 0.0156,
       "step": 8677
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.898320808980425e-06,
-      "loss": 0.1381,
+      "loss": 0.0236,
       "step": 8678
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.896465349290287e-06,
-      "loss": 0.3005,
+      "loss": 0.083,
       "step": 8679
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.894609889600149e-06,
-      "loss": 0.3585,
+      "loss": 0.1074,
       "step": 8680
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.892754429910011e-06,
-      "loss": 0.1361,
+      "loss": 0.0818,
       "step": 8681
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.8908989702198726e-06,
-      "loss": 0.1416,
+      "loss": 0.0189,
       "step": 8682
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.8890435105297344e-06,
-      "loss": 0.1404,
+      "loss": 0.0871,
       "step": 8683
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.887188050839596e-06,
-      "loss": 0.1903,
+      "loss": 0.0633,
       "step": 8684
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.885332591149457e-06,
-      "loss": 0.1861,
+      "loss": 0.1676,
       "step": 8685
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.883477131459319e-06,
-      "loss": 0.2956,
+      "loss": 0.0516,
       "step": 8686
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.881621671769181e-06,
-      "loss": 0.1882,
+      "loss": 0.0628,
       "step": 8687
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.879766212079043e-06,
-      "loss": 0.086,
+      "loss": 0.0212,
       "step": 8688
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.877910752388905e-06,
-      "loss": 0.1898,
+      "loss": 0.1193,
       "step": 8689
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.876055292698767e-06,
-      "loss": 0.1369,
+      "loss": 0.0594,
       "step": 8690
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.8741998330086285e-06,
-      "loss": 0.0905,
+      "loss": 0.0734,
       "step": 8691
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.87234437331849e-06,
-      "loss": 0.2416,
+      "loss": 0.1403,
       "step": 8692
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.870488913628352e-06,
-      "loss": 0.1885,
+      "loss": 0.0284,
       "step": 8693
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.868633453938213e-06,
-      "loss": 0.0871,
+      "loss": 0.0747,
       "step": 8694
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.866777994248075e-06,
-      "loss": 0.0872,
+      "loss": 0.0835,
       "step": 8695
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.864922534557937e-06,
-      "loss": 0.0356,
+      "loss": 0.0759,
       "step": 8696
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.863067074867799e-06,
-      "loss": 0.087,
+      "loss": 0.1006,
       "step": 8697
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.861211615177661e-06,
-      "loss": 0.0902,
+      "loss": 0.0504,
       "step": 8698
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.859356155487523e-06,
-      "loss": 0.1395,
+      "loss": 0.1171,
       "step": 8699
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.8575006957973845e-06,
-      "loss": 0.1925,
+      "loss": 0.0735,
       "step": 8700
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.855645236107246e-06,
-      "loss": 0.1408,
+      "loss": 0.0165,
       "step": 8701
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.8537897764171074e-06,
-      "loss": 0.3478,
+      "loss": 0.0187,
       "step": 8702
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.851934316726969e-06,
-      "loss": 0.4021,
+      "loss": 0.0182,
       "step": 8703
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.850078857036831e-06,
-      "loss": 0.0897,
+      "loss": 0.054,
       "step": 8704
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.848223397346693e-06,
-      "loss": 0.0876,
+      "loss": 0.047,
       "step": 8705
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.846367937656555e-06,
-      "loss": 0.1915,
+      "loss": 0.0189,
       "step": 8706
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.844512477966417e-06,
-      "loss": 0.1933,
+      "loss": 0.0509,
       "step": 8707
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.842657018276279e-06,
-      "loss": 0.1921,
+      "loss": 0.0537,
       "step": 8708
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.8408015585861405e-06,
-      "loss": 0.291,
+      "loss": 0.0933,
       "step": 8709
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.838946098896002e-06,
-      "loss": 0.2441,
+      "loss": 0.0812,
       "step": 8710
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.837090639205863e-06,
-      "loss": 0.1919,
+      "loss": 0.0529,
       "step": 8711
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.835235179515725e-06,
-      "loss": 0.1874,
+      "loss": 0.1564,
       "step": 8712
     },
     {
       "epoch": 2.42,
       "learning_rate": 3.833379719825587e-06,
-      "loss": 0.1382,
+      "loss": 0.0131,
       "step": 8713
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.831524260135449e-06,
-      "loss": 0.3974,
+      "loss": 0.0512,
       "step": 8714
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.829668800445311e-06,
-      "loss": 0.0907,
+      "loss": 0.0157,
       "step": 8715
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.827813340755173e-06,
-      "loss": 0.287,
+      "loss": 0.0653,
       "step": 8716
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.825957881065035e-06,
-      "loss": 0.1379,
+      "loss": 0.0953,
       "step": 8717
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.8241024213748965e-06,
-      "loss": 0.2898,
+      "loss": 0.0854,
       "step": 8718
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.8222469616847575e-06,
-      "loss": 0.091,
+      "loss": 0.0831,
       "step": 8719
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.820391501994619e-06,
-      "loss": 0.1404,
+      "loss": 0.0547,
       "step": 8720
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.818536042304481e-06,
-      "loss": 0.1913,
+      "loss": 0.0811,
       "step": 8721
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.816680582614343e-06,
-      "loss": 0.1945,
+      "loss": 0.1104,
       "step": 8722
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.814825122924205e-06,
-      "loss": 0.1871,
+      "loss": 0.0213,
       "step": 8723
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.812969663234067e-06,
-      "loss": 0.1862,
+      "loss": 0.0735,
       "step": 8724
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.8111142035439288e-06,
-      "loss": 0.0932,
+      "loss": 0.0955,
       "step": 8725
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.80925874385379e-06,
-      "loss": 0.0925,
+      "loss": 0.0572,
       "step": 8726
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.807403284163652e-06,
-      "loss": 0.1425,
+      "loss": 0.1588,
       "step": 8727
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.805547824473514e-06,
-      "loss": 0.1888,
+      "loss": 0.0432,
       "step": 8728
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.803692364783376e-06,
-      "loss": 0.1859,
+      "loss": 0.0639,
       "step": 8729
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.801836905093237e-06,
-      "loss": 0.0923,
+      "loss": 0.0472,
       "step": 8730
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.7999814454030987e-06,
-      "loss": 0.1403,
+      "loss": 0.0189,
       "step": 8731
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.7981259857129606e-06,
-      "loss": 0.1412,
+      "loss": 0.0565,
       "step": 8732
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.796270526022822e-06,
-      "loss": 0.1907,
+      "loss": 0.0696,
       "step": 8733
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.794415066332684e-06,
-      "loss": 0.142,
+      "loss": 0.0534,
       "step": 8734
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.7925596066425458e-06,
-      "loss": 0.1871,
+      "loss": 0.0233,
       "step": 8735
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.7907041469524076e-06,
-      "loss": 0.1437,
+      "loss": 0.0593,
       "step": 8736
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.788848687262269e-06,
-      "loss": 0.192,
+      "loss": 0.0205,
       "step": 8737
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.786993227572131e-06,
-      "loss": 0.0888,
+      "loss": 0.1029,
       "step": 8738
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.785137767881993e-06,
-      "loss": 0.1392,
+      "loss": 0.0685,
       "step": 8739
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.7832823081918547e-06,
-      "loss": 0.1422,
+      "loss": 0.0433,
       "step": 8740
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.7814268485017166e-06,
-      "loss": 0.1907,
+      "loss": 0.0201,
       "step": 8741
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.779571388811578e-06,
-      "loss": 0.1898,
+      "loss": 0.016,
       "step": 8742
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.77771592912144e-06,
-      "loss": 0.042,
+      "loss": 0.1929,
       "step": 8743
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.7758604694313018e-06,
-      "loss": 0.0908,
+      "loss": 0.022,
       "step": 8744
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.7740050097411636e-06,
-      "loss": 0.0883,
+      "loss": 0.0889,
       "step": 8745
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.772149550051025e-06,
-      "loss": 0.041,
+      "loss": 0.0617,
       "step": 8746
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.770294090360887e-06,
-      "loss": 0.0394,
+      "loss": 0.0652,
       "step": 8747
     },
     {
       "epoch": 2.43,
       "learning_rate": 3.768438630670749e-06,
-      "loss": 0.0389,
+      "loss": 0.0625,
       "step": 8748
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.7665831709806107e-06,
-      "loss": 0.0899,
+      "loss": 0.0892,
       "step": 8749
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.764727711290472e-06,
-      "loss": 0.139,
+      "loss": 0.0226,
       "step": 8750
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.762872251600334e-06,
-      "loss": 0.1921,
+      "loss": 0.0089,
       "step": 8751
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.761016791910196e-06,
-      "loss": 0.0903,
+      "loss": 0.0353,
       "step": 8752
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.7591613322200577e-06,
-      "loss": 0.1383,
+      "loss": 0.0897,
       "step": 8753
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.7573058725299196e-06,
-      "loss": 0.1926,
+      "loss": 0.1194,
       "step": 8754
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.755450412839781e-06,
-      "loss": 0.0889,
+      "loss": 0.0494,
       "step": 8755
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.753594953149643e-06,
-      "loss": 0.2429,
+      "loss": 0.0712,
       "step": 8756
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.751739493459505e-06,
-      "loss": 0.1864,
+      "loss": 0.0831,
       "step": 8757
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.7498840337693667e-06,
-      "loss": 0.0353,
+      "loss": 0.0266,
       "step": 8758
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.748028574079228e-06,
-      "loss": 0.1398,
+      "loss": 0.1537,
       "step": 8759
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.74617311438909e-06,
-      "loss": 0.4002,
+      "loss": 0.038,
       "step": 8760
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.744317654698952e-06,
-      "loss": 0.1899,
+      "loss": 0.014,
       "step": 8761
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.7424621950088137e-06,
-      "loss": 0.0869,
+      "loss": 0.0177,
       "step": 8762
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.740606735318675e-06,
-      "loss": 0.3422,
+      "loss": 0.0146,
       "step": 8763
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.738751275628537e-06,
-      "loss": 0.1859,
+      "loss": 0.0886,
       "step": 8764
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.736895815938399e-06,
-      "loss": 0.2434,
+      "loss": 0.1975,
       "step": 8765
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.735040356248261e-06,
-      "loss": 0.0868,
+      "loss": 0.1744,
       "step": 8766
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.7331848965581222e-06,
-      "loss": 0.1377,
+      "loss": 0.0141,
       "step": 8767
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.731329436867984e-06,
-      "loss": 0.2432,
+      "loss": 0.0217,
       "step": 8768
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.729473977177846e-06,
-      "loss": 0.296,
+      "loss": 0.0619,
       "step": 8769
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.727618517487708e-06,
-      "loss": 0.1888,
+      "loss": 0.0574,
       "step": 8770
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.7257630577975697e-06,
-      "loss": 0.1424,
+      "loss": 0.0948,
       "step": 8771
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.723907598107431e-06,
-      "loss": 0.1375,
+      "loss": 0.0161,
       "step": 8772
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.722052138417293e-06,
-      "loss": 0.0902,
+      "loss": 0.0859,
       "step": 8773
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.720196678727155e-06,
-      "loss": 0.1873,
+      "loss": 0.1547,
       "step": 8774
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.7183412190370168e-06,
-      "loss": 0.0874,
+      "loss": 0.146,
       "step": 8775
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.7164857593468782e-06,
-      "loss": 0.1914,
+      "loss": 0.1288,
       "step": 8776
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.71463029965674e-06,
-      "loss": 0.0883,
+      "loss": 0.0138,
       "step": 8777
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.712774839966602e-06,
-      "loss": 0.1929,
+      "loss": 0.1129,
       "step": 8778
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.710919380276464e-06,
-      "loss": 0.2408,
+      "loss": 0.1301,
       "step": 8779
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.7090639205863253e-06,
-      "loss": 0.0877,
+      "loss": 0.0186,
       "step": 8780
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.707208460896187e-06,
-      "loss": 0.188,
+      "loss": 0.0907,
       "step": 8781
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.705353001206049e-06,
-      "loss": 0.0875,
+      "loss": 0.0228,
       "step": 8782
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.703497541515911e-06,
-      "loss": 0.2443,
+      "loss": 0.0178,
       "step": 8783
     },
     {
       "epoch": 2.44,
       "learning_rate": 3.7016420818257728e-06,
-      "loss": 0.3446,
+      "loss": 0.0399,
       "step": 8784
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6997866221356342e-06,
-      "loss": 0.1405,
+      "loss": 0.0678,
       "step": 8785
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.697931162445496e-06,
-      "loss": 0.0381,
+      "loss": 0.0421,
       "step": 8786
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.696075702755358e-06,
-      "loss": 0.1906,
+      "loss": 0.0244,
       "step": 8787
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.69422024306522e-06,
-      "loss": 0.0885,
+      "loss": 0.1183,
       "step": 8788
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6923647833750813e-06,
-      "loss": 0.1384,
+      "loss": 0.0611,
       "step": 8789
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.690509323684943e-06,
-      "loss": 0.0887,
+      "loss": 0.1019,
       "step": 8790
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.688653863994805e-06,
-      "loss": 0.0369,
+      "loss": 0.1376,
       "step": 8791
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.686798404304667e-06,
-      "loss": 0.2447,
+      "loss": 0.1075,
       "step": 8792
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6849429446145283e-06,
-      "loss": 0.0906,
+      "loss": 0.0239,
       "step": 8793
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.68308748492439e-06,
-      "loss": 0.1903,
+      "loss": 0.0442,
       "step": 8794
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.681232025234252e-06,
-      "loss": 0.1466,
+      "loss": 0.1435,
       "step": 8795
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.679376565544114e-06,
-      "loss": 0.1394,
+      "loss": 0.0129,
       "step": 8796
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6775211058539754e-06,
-      "loss": 0.1442,
+      "loss": 0.0193,
       "step": 8797
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6756656461638373e-06,
-      "loss": 0.0881,
+      "loss": 0.0234,
       "step": 8798
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.673810186473699e-06,
-      "loss": 0.1896,
+      "loss": 0.0674,
       "step": 8799
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.671954726783561e-06,
-      "loss": 0.1424,
+      "loss": 0.0274,
       "step": 8800
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.670099267093423e-06,
-      "loss": 0.0899,
+      "loss": 0.0589,
       "step": 8801
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6682438074032843e-06,
-      "loss": 0.0882,
+      "loss": 0.0221,
       "step": 8802
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.666388347713146e-06,
-      "loss": 0.1912,
+      "loss": 0.0492,
       "step": 8803
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.664532888023008e-06,
-      "loss": 0.0875,
+      "loss": 0.0254,
       "step": 8804
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.66267742833287e-06,
-      "loss": 0.1397,
+      "loss": 0.1057,
       "step": 8805
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6608219686427314e-06,
-      "loss": 0.1389,
+      "loss": 0.014,
       "step": 8806
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6589665089525932e-06,
-      "loss": 0.1357,
+      "loss": 0.0206,
       "step": 8807
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.657111049262455e-06,
-      "loss": 0.2423,
+      "loss": 0.0481,
       "step": 8808
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.655255589572317e-06,
-      "loss": 0.1928,
+      "loss": 0.0147,
       "step": 8809
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6534001298821784e-06,
-      "loss": 0.1909,
+      "loss": 0.0935,
       "step": 8810
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6515446701920403e-06,
-      "loss": 0.138,
+      "loss": 0.1457,
       "step": 8811
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.649689210501902e-06,
-      "loss": 0.0888,
+      "loss": 0.0164,
       "step": 8812
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.647833750811764e-06,
-      "loss": 0.0886,
+      "loss": 0.1316,
       "step": 8813
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.645978291121626e-06,
-      "loss": 0.0865,
+      "loss": 0.06,
       "step": 8814
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6441228314314874e-06,
-      "loss": 0.1892,
+      "loss": 0.1239,
       "step": 8815
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6422673717413492e-06,
-      "loss": 0.0347,
+      "loss": 0.1323,
       "step": 8816
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.640411912051211e-06,
-      "loss": 0.0863,
+      "loss": 0.1427,
       "step": 8817
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.638556452361073e-06,
-      "loss": 0.0353,
+      "loss": 0.1423,
       "step": 8818
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6367009926709344e-06,
-      "loss": 0.1941,
+      "loss": 0.0169,
       "step": 8819
     },
     {
       "epoch": 2.45,
       "learning_rate": 3.6348455329807963e-06,
-      "loss": 0.0885,
+      "loss": 0.0763,
       "step": 8820
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.632990073290658e-06,
-      "loss": 0.1402,
+      "loss": 0.1266,
       "step": 8821
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.63113461360052e-06,
-      "loss": 0.0858,
+      "loss": 0.1041,
       "step": 8822
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.6292791539103815e-06,
-      "loss": 0.1401,
+      "loss": 0.0715,
       "step": 8823
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.6274236942202434e-06,
-      "loss": 0.192,
+      "loss": 0.0405,
       "step": 8824
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.6255682345301052e-06,
-      "loss": 0.2984,
+      "loss": 0.0773,
       "step": 8825
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.623712774839967e-06,
-      "loss": 0.242,
+      "loss": 0.0216,
       "step": 8826
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.6218573151498285e-06,
-      "loss": 0.1375,
+      "loss": 0.0086,
       "step": 8827
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.6200018554596904e-06,
-      "loss": 0.1379,
+      "loss": 0.1049,
       "step": 8828
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.6181463957695523e-06,
-      "loss": 0.196,
+      "loss": 0.1002,
       "step": 8829
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.616290936079414e-06,
-      "loss": 0.0884,
+      "loss": 0.0617,
       "step": 8830
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.614435476389276e-06,
-      "loss": 0.0334,
+      "loss": 0.09,
       "step": 8831
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.6125800166991375e-06,
-      "loss": 0.2444,
+      "loss": 0.0979,
       "step": 8832
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.6107245570089993e-06,
-      "loss": 0.2461,
+      "loss": 0.0161,
       "step": 8833
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.6088690973188612e-06,
-      "loss": 0.0859,
+      "loss": 0.0217,
       "step": 8834
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.607013637628723e-06,
-      "loss": 0.0846,
+      "loss": 0.0756,
       "step": 8835
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.6051581779385845e-06,
-      "loss": 0.2455,
+      "loss": 0.0883,
       "step": 8836
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.6033027182484464e-06,
-      "loss": 0.2431,
+      "loss": 0.0979,
       "step": 8837
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.6014472585583083e-06,
-      "loss": 0.1913,
+      "loss": 0.0964,
       "step": 8838
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.59959179886817e-06,
-      "loss": 0.087,
+      "loss": 0.1084,
       "step": 8839
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5977363391780316e-06,
-      "loss": 0.1408,
+      "loss": 0.0261,
       "step": 8840
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5958808794878935e-06,
-      "loss": 0.2465,
+      "loss": 0.0901,
       "step": 8841
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5940254197977553e-06,
-      "loss": 0.0874,
+      "loss": 0.062,
       "step": 8842
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.592169960107617e-06,
-      "loss": 0.1909,
+      "loss": 0.077,
       "step": 8843
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.590314500417479e-06,
-      "loss": 0.0337,
+      "loss": 0.0179,
       "step": 8844
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5884590407273405e-06,
-      "loss": 0.083,
+      "loss": 0.0845,
       "step": 8845
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5866035810372024e-06,
-      "loss": 0.0852,
+      "loss": 0.0182,
       "step": 8846
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5847481213470643e-06,
-      "loss": 0.1362,
+      "loss": 0.0958,
       "step": 8847
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.582892661656926e-06,
-      "loss": 0.1907,
+      "loss": 0.0452,
       "step": 8848
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5810372019667876e-06,
-      "loss": 0.1876,
+      "loss": 0.0457,
       "step": 8849
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5791817422766494e-06,
-      "loss": 0.297,
+      "loss": 0.0439,
       "step": 8850
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5773262825865113e-06,
-      "loss": 0.2431,
+      "loss": 0.0651,
       "step": 8851
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.575470822896373e-06,
-      "loss": 0.2457,
+      "loss": 0.0166,
       "step": 8852
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5736153632062346e-06,
-      "loss": 0.1409,
+      "loss": 0.018,
       "step": 8853
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5717599035160965e-06,
-      "loss": 0.0348,
+      "loss": 0.0476,
       "step": 8854
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5699044438259584e-06,
-      "loss": 0.1894,
+      "loss": 0.0132,
       "step": 8855
     },
     {
       "epoch": 2.46,
       "learning_rate": 3.5680489841358202e-06,
-      "loss": 0.0876,
+      "loss": 0.0622,
       "step": 8856
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5661935244456817e-06,
-      "loss": 0.1931,
+      "loss": 0.0287,
       "step": 8857
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5643380647555436e-06,
-      "loss": 0.0878,
+      "loss": 0.0938,
       "step": 8858
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5624826050654054e-06,
-      "loss": 0.3928,
+      "loss": 0.0641,
       "step": 8859
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5606271453752673e-06,
-      "loss": 0.2958,
+      "loss": 0.032,
       "step": 8860
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.558771685685129e-06,
-      "loss": 0.0888,
+      "loss": 0.0605,
       "step": 8861
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5569162259949906e-06,
-      "loss": 0.0357,
+      "loss": 0.0185,
       "step": 8862
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5550607663048525e-06,
-      "loss": 0.1884,
+      "loss": 0.0488,
       "step": 8863
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5532053066147144e-06,
-      "loss": 0.1406,
+      "loss": 0.0928,
       "step": 8864
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5513498469245762e-06,
-      "loss": 0.0346,
+      "loss": 0.0566,
       "step": 8865
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5494943872344377e-06,
-      "loss": 0.2935,
+      "loss": 0.0291,
       "step": 8866
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5476389275442996e-06,
-      "loss": 0.1417,
+      "loss": 0.1008,
       "step": 8867
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5457834678541614e-06,
-      "loss": 0.0881,
+      "loss": 0.0146,
       "step": 8868
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5439280081640233e-06,
-      "loss": 0.2442,
+      "loss": 0.052,
       "step": 8869
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5420725484738847e-06,
-      "loss": 0.036,
+      "loss": 0.0994,
       "step": 8870
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5402170887837466e-06,
-      "loss": 0.1357,
+      "loss": 0.0154,
       "step": 8871
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5383616290936085e-06,
-      "loss": 0.1407,
+      "loss": 0.0992,
       "step": 8872
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5365061694034704e-06,
-      "loss": 0.1918,
+      "loss": 0.0106,
       "step": 8873
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5346507097133322e-06,
-      "loss": 0.1943,
+      "loss": 0.0535,
       "step": 8874
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5327952500231937e-06,
-      "loss": 0.1421,
+      "loss": 0.1507,
       "step": 8875
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5309397903330555e-06,
-      "loss": 0.0875,
+      "loss": 0.0652,
       "step": 8876
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5290843306429174e-06,
-      "loss": 0.0868,
+      "loss": 0.0195,
       "step": 8877
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5272288709527793e-06,
-      "loss": 0.0374,
+      "loss": 0.2295,
       "step": 8878
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5253734112626407e-06,
-      "loss": 0.089,
+      "loss": 0.0836,
       "step": 8879
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5235179515725026e-06,
-      "loss": 0.0892,
+      "loss": 0.0512,
       "step": 8880
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5216624918823645e-06,
-      "loss": 0.1927,
+      "loss": 0.0498,
       "step": 8881
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5198070321922255e-06,
-      "loss": 0.2429,
+      "loss": 0.0488,
       "step": 8882
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5179515725020874e-06,
-      "loss": 0.0858,
+      "loss": 0.0713,
       "step": 8883
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5160961128119492e-06,
-      "loss": 0.2426,
+      "loss": 0.0659,
       "step": 8884
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.514240653121811e-06,
-      "loss": 0.1933,
+      "loss": 0.0405,
       "step": 8885
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5123851934316726e-06,
-      "loss": 0.188,
+      "loss": 0.0594,
       "step": 8886
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5105297337415344e-06,
-      "loss": 0.1958,
+      "loss": 0.1038,
       "step": 8887
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5086742740513963e-06,
-      "loss": 0.2424,
+      "loss": 0.1327,
       "step": 8888
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.506818814361258e-06,
-      "loss": 0.1352,
+      "loss": 0.049,
       "step": 8889
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5049633546711196e-06,
-      "loss": 0.1387,
+      "loss": 0.0212,
       "step": 8890
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5031078949809815e-06,
-      "loss": 0.1938,
+      "loss": 0.1415,
       "step": 8891
     },
     {
       "epoch": 2.47,
       "learning_rate": 3.5012524352908434e-06,
-      "loss": 0.0859,
+      "loss": 0.1567,
       "step": 8892
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4993969756007052e-06,
-      "loss": 0.191,
+      "loss": 0.016,
       "step": 8893
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.497541515910567e-06,
-      "loss": 0.1364,
+      "loss": 0.0197,
       "step": 8894
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4956860562204285e-06,
-      "loss": 0.0875,
+      "loss": 0.1397,
       "step": 8895
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4938305965302904e-06,
-      "loss": 0.1913,
+      "loss": 0.056,
       "step": 8896
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4919751368401523e-06,
-      "loss": 0.0879,
+      "loss": 0.1123,
       "step": 8897
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.490119677150014e-06,
-      "loss": 0.0888,
+      "loss": 0.0567,
       "step": 8898
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4882642174598756e-06,
-      "loss": 0.0866,
+      "loss": 0.0197,
       "step": 8899
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4864087577697375e-06,
-      "loss": 0.143,
+      "loss": 0.0237,
       "step": 8900
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4845532980795993e-06,
-      "loss": 0.086,
+      "loss": 0.0128,
       "step": 8901
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4826978383894612e-06,
-      "loss": 0.1918,
+      "loss": 0.0119,
       "step": 8902
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4808423786993227e-06,
-      "loss": 0.2934,
+      "loss": 0.0204,
       "step": 8903
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4789869190091845e-06,
-      "loss": 0.035,
+      "loss": 0.0477,
       "step": 8904
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4771314593190464e-06,
-      "loss": 0.24,
+      "loss": 0.1017,
       "step": 8905
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4752759996289083e-06,
-      "loss": 0.2392,
+      "loss": 0.0953,
       "step": 8906
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.47342053993877e-06,
-      "loss": 0.1401,
+      "loss": 0.0254,
       "step": 8907
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4715650802486316e-06,
-      "loss": 0.0901,
+      "loss": 0.0613,
       "step": 8908
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4697096205584935e-06,
-      "loss": 0.0872,
+      "loss": 0.0533,
       "step": 8909
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4678541608683553e-06,
-      "loss": 0.1986,
+      "loss": 0.0602,
       "step": 8910
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.465998701178217e-06,
-      "loss": 0.0889,
+      "loss": 0.0581,
       "step": 8911
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4641432414880786e-06,
-      "loss": 0.0877,
+      "loss": 0.0166,
       "step": 8912
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4622877817979405e-06,
-      "loss": 0.2982,
+      "loss": 0.0578,
       "step": 8913
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4604323221078024e-06,
-      "loss": 0.2497,
+      "loss": 0.0098,
       "step": 8914
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4585768624176643e-06,
-      "loss": 0.1412,
+      "loss": 0.0457,
       "step": 8915
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4567214027275257e-06,
-      "loss": 0.2415,
+      "loss": 0.1405,
       "step": 8916
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4548659430373876e-06,
-      "loss": 0.0369,
+      "loss": 0.1408,
       "step": 8917
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4530104833472494e-06,
-      "loss": 0.0347,
+      "loss": 0.0119,
       "step": 8918
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4511550236571113e-06,
-      "loss": 0.0893,
+      "loss": 0.0681,
       "step": 8919
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4492995639669728e-06,
-      "loss": 0.1913,
+      "loss": 0.0096,
       "step": 8920
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4474441042768346e-06,
-      "loss": 0.0885,
+      "loss": 0.0143,
       "step": 8921
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4455886445866965e-06,
-      "loss": 0.2404,
+      "loss": 0.1264,
       "step": 8922
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4437331848965584e-06,
-      "loss": 0.1373,
+      "loss": 0.0959,
       "step": 8923
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4418777252064202e-06,
-      "loss": 0.0885,
+      "loss": 0.0821,
       "step": 8924
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4400222655162817e-06,
-      "loss": 0.1925,
+      "loss": 0.169,
       "step": 8925
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4381668058261436e-06,
-      "loss": 0.1373,
+      "loss": 0.0158,
       "step": 8926
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4363113461360054e-06,
-      "loss": 0.0876,
+      "loss": 0.1176,
       "step": 8927
     },
     {
       "epoch": 2.48,
       "learning_rate": 3.4344558864458673e-06,
-      "loss": 0.1403,
+      "loss": 0.1221,
       "step": 8928
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4326004267557288e-06,
-      "loss": 0.1381,
+      "loss": 0.0466,
       "step": 8929
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4307449670655906e-06,
-      "loss": 0.1875,
+      "loss": 0.022,
       "step": 8930
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4288895073754525e-06,
-      "loss": 0.1934,
+      "loss": 0.0216,
       "step": 8931
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4270340476853144e-06,
-      "loss": 0.1392,
+      "loss": 0.0703,
       "step": 8932
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.425178587995176e-06,
-      "loss": 0.0878,
+      "loss": 0.1877,
       "step": 8933
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4233231283050377e-06,
-      "loss": 0.1392,
+      "loss": 0.0368,
       "step": 8934
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4214676686148996e-06,
-      "loss": 0.1404,
+      "loss": 0.0851,
       "step": 8935
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4196122089247614e-06,
-      "loss": 0.1899,
+      "loss": 0.1009,
       "step": 8936
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4177567492346233e-06,
-      "loss": 0.0864,
+      "loss": 0.1433,
       "step": 8937
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4159012895444847e-06,
-      "loss": 0.0348,
+      "loss": 0.0191,
       "step": 8938
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4140458298543466e-06,
-      "loss": 0.1891,
+      "loss": 0.0939,
       "step": 8939
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4121903701642085e-06,
-      "loss": 0.0855,
+      "loss": 0.0505,
       "step": 8940
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4103349104740704e-06,
-      "loss": 0.1389,
+      "loss": 0.1138,
       "step": 8941
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.408479450783932e-06,
-      "loss": 0.0874,
+      "loss": 0.044,
       "step": 8942
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4066239910937937e-06,
-      "loss": 0.0843,
+      "loss": 0.0192,
       "step": 8943
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4047685314036555e-06,
-      "loss": 0.0855,
+      "loss": 0.0132,
       "step": 8944
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.4029130717135174e-06,
-      "loss": 0.0335,
+      "loss": 0.0631,
       "step": 8945
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.401057612023379e-06,
-      "loss": 0.3509,
+      "loss": 0.1145,
       "step": 8946
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.3992021523332407e-06,
-      "loss": 0.0869,
+      "loss": 0.014,
       "step": 8947
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.3973466926431026e-06,
-      "loss": 0.1389,
+      "loss": 0.0984,
       "step": 8948
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.3954912329529645e-06,
-      "loss": 0.0858,
+      "loss": 0.0125,
       "step": 8949
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.393635773262826e-06,
-      "loss": 0.3025,
+      "loss": 0.0774,
       "step": 8950
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.391780313572688e-06,
-      "loss": 0.141,
+      "loss": 0.1035,
       "step": 8951
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.3899248538825497e-06,
-      "loss": 0.1391,
+      "loss": 0.1493,
       "step": 8952
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.3880693941924115e-06,
-      "loss": 0.0864,
+      "loss": 0.1204,
       "step": 8953
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.3862139345022734e-06,
-      "loss": 0.193,
+      "loss": 0.0444,
       "step": 8954
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.384358474812135e-06,
-      "loss": 0.2394,
+      "loss": 0.0153,
       "step": 8955
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.3825030151219967e-06,
-      "loss": 0.0336,
+      "loss": 0.0609,
       "step": 8956
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.3806475554318586e-06,
-      "loss": 0.1386,
+      "loss": 0.1247,
       "step": 8957
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.3787920957417205e-06,
-      "loss": 0.0345,
+      "loss": 0.0721,
       "step": 8958
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.376936636051582e-06,
-      "loss": 0.1911,
+      "loss": 0.0414,
       "step": 8959
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.3750811763614438e-06,
-      "loss": 0.0854,
+      "loss": 0.0201,
       "step": 8960
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.3732257166713056e-06,
-      "loss": 0.1376,
+      "loss": 0.0237,
       "step": 8961
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.3713702569811675e-06,
-      "loss": 0.1396,
+      "loss": 0.0972,
       "step": 8962
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.369514797291029e-06,
-      "loss": 0.0835,
+      "loss": 0.1121,
       "step": 8963
     },
     {
       "epoch": 2.49,
       "learning_rate": 3.367659337600891e-06,
-      "loss": 0.139,
+      "loss": 0.0963,
       "step": 8964
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3658038779107527e-06,
-      "loss": 0.1418,
+      "loss": 0.0174,
       "step": 8965
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3639484182206146e-06,
-      "loss": 0.2966,
+      "loss": 0.0561,
       "step": 8966
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3620929585304764e-06,
-      "loss": 0.1902,
+      "loss": 0.024,
       "step": 8967
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.360237498840338e-06,
-      "loss": 0.2483,
+      "loss": 0.1054,
       "step": 8968
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3583820391501998e-06,
-      "loss": 0.1389,
+      "loss": 0.017,
       "step": 8969
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3565265794600616e-06,
-      "loss": 0.1851,
+      "loss": 0.0774,
       "step": 8970
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3546711197699235e-06,
-      "loss": 0.0325,
+      "loss": 0.0643,
       "step": 8971
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.352815660079785e-06,
-      "loss": 0.2977,
+      "loss": 0.1674,
       "step": 8972
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.350960200389647e-06,
-      "loss": 0.0337,
+      "loss": 0.0791,
       "step": 8973
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3491047406995087e-06,
-      "loss": 0.136,
+      "loss": 0.0894,
       "step": 8974
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3472492810093706e-06,
-      "loss": 0.0861,
+      "loss": 0.3238,
       "step": 8975
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.345393821319232e-06,
-      "loss": 0.1359,
+      "loss": 0.16,
       "step": 8976
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.343538361629094e-06,
-      "loss": 0.1418,
+      "loss": 0.0219,
       "step": 8977
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3416829019389558e-06,
-      "loss": 0.1942,
+      "loss": 0.0476,
       "step": 8978
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3398274422488176e-06,
-      "loss": 0.1879,
+      "loss": 0.0892,
       "step": 8979
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.337971982558679e-06,
-      "loss": 0.137,
+      "loss": 0.0903,
       "step": 8980
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.336116522868541e-06,
-      "loss": 0.1928,
+      "loss": 0.1342,
       "step": 8981
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.334261063178403e-06,
-      "loss": 0.1394,
+      "loss": 0.0222,
       "step": 8982
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3324056034882647e-06,
-      "loss": 0.1892,
+      "loss": 0.0699,
       "step": 8983
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3305501437981266e-06,
-      "loss": 0.1371,
+      "loss": 0.0517,
       "step": 8984
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.328694684107988e-06,
-      "loss": 0.1916,
+      "loss": 0.194,
       "step": 8985
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.32683922441785e-06,
-      "loss": 0.0341,
+      "loss": 0.048,
       "step": 8986
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3249837647277117e-06,
-      "loss": 0.0869,
+      "loss": 0.0207,
       "step": 8987
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3231283050375736e-06,
-      "loss": 0.0854,
+      "loss": 0.0235,
       "step": 8988
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.321272845347435e-06,
-      "loss": 0.1375,
+      "loss": 0.0173,
       "step": 8989
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.319417385657297e-06,
-      "loss": 0.1922,
+      "loss": 0.0967,
       "step": 8990
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.317561925967159e-06,
-      "loss": 0.1952,
+      "loss": 0.0913,
       "step": 8991
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3157064662770207e-06,
-      "loss": 0.1352,
+      "loss": 0.0266,
       "step": 8992
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.313851006586882e-06,
-      "loss": 0.1931,
+      "loss": 0.0274,
       "step": 8993
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.311995546896744e-06,
-      "loss": 0.1397,
+      "loss": 0.0221,
       "step": 8994
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.310140087206606e-06,
-      "loss": 0.1907,
+      "loss": 0.23,
       "step": 8995
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3082846275164677e-06,
-      "loss": 0.1373,
+      "loss": 0.0258,
       "step": 8996
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3064291678263296e-06,
-      "loss": 0.1906,
+      "loss": 0.1201,
       "step": 8997
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.304573708136191e-06,
-      "loss": 0.1906,
+      "loss": 0.0455,
       "step": 8998
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.302718248446053e-06,
-      "loss": 0.0863,
+      "loss": 0.0159,
       "step": 8999
     },
     {
       "epoch": 2.5,
       "learning_rate": 3.3008627887559148e-06,
-      "loss": 0.0868,
+      "loss": 0.0179,
       "step": 9000
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.2990073290657767e-06,
-      "loss": 0.249,
+      "loss": 0.0771,
       "step": 9001
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.297151869375638e-06,
-      "loss": 0.2413,
+      "loss": 0.0545,
       "step": 9002
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.2952964096855e-06,
-      "loss": 0.1378,
+      "loss": 0.056,
       "step": 9003
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.293440949995362e-06,
-      "loss": 0.1409,
+      "loss": 0.1444,
       "step": 9004
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.2915854903052237e-06,
-      "loss": 0.0888,
+      "loss": 0.0191,
       "step": 9005
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.289730030615085e-06,
-      "loss": 0.0826,
+      "loss": 0.0667,
       "step": 9006
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.287874570924947e-06,
-      "loss": 0.0862,
+      "loss": 0.0377,
       "step": 9007
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.286019111234809e-06,
-      "loss": 0.0339,
+      "loss": 0.0123,
       "step": 9008
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.2841636515446708e-06,
-      "loss": 0.2958,
+      "loss": 0.1079,
       "step": 9009
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.2823081918545322e-06,
-      "loss": 0.1937,
+      "loss": 0.0224,
       "step": 9010
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.280452732164394e-06,
-      "loss": 0.0881,
+      "loss": 0.1427,
       "step": 9011
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.278597272474256e-06,
-      "loss": 0.193,
+      "loss": 0.0599,
       "step": 9012
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.276741812784118e-06,
-      "loss": 0.0885,
+      "loss": 0.044,
       "step": 9013
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.2748863530939797e-06,
-      "loss": 0.0345,
+      "loss": 0.1236,
       "step": 9014
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.273030893403841e-06,
-      "loss": 0.1928,
+      "loss": 0.1622,
       "step": 9015
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.271175433713703e-06,
-      "loss": 0.1918,
+      "loss": 0.0705,
       "step": 9016
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.269319974023565e-06,
-      "loss": 0.1926,
+      "loss": 0.0164,
       "step": 9017
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.2674645143334268e-06,
-      "loss": 0.0346,
+      "loss": 0.0849,
       "step": 9018
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.265609054643288e-06,
-      "loss": 0.1916,
+      "loss": 0.1252,
       "step": 9019
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.26375359495315e-06,
-      "loss": 0.1913,
+      "loss": 0.0822,
       "step": 9020
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.261898135263012e-06,
-      "loss": 0.0857,
+      "loss": 0.0503,
       "step": 9021
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.260042675572874e-06,
-      "loss": 0.1406,
+      "loss": 0.0161,
       "step": 9022
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.2581872158827353e-06,
-      "loss": 0.0338,
+      "loss": 0.038,
       "step": 9023
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.256331756192597e-06,
-      "loss": 0.0873,
+      "loss": 0.1131,
       "step": 9024
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.254476296502459e-06,
-      "loss": 0.1396,
+      "loss": 0.2051,
       "step": 9025
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.252620836812321e-06,
-      "loss": 0.144,
+      "loss": 0.062,
       "step": 9026
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.2507653771221828e-06,
-      "loss": 0.192,
+      "loss": 0.0166,
       "step": 9027
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.248909917432044e-06,
-      "loss": 0.0339,
+      "loss": 0.0153,
       "step": 9028
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.247054457741906e-06,
-      "loss": 0.0865,
+      "loss": 0.1714,
       "step": 9029
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.245198998051768e-06,
-      "loss": 0.1416,
+      "loss": 0.0534,
       "step": 9030
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.24334353836163e-06,
-      "loss": 0.0844,
+      "loss": 0.1145,
       "step": 9031
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.2414880786714913e-06,
-      "loss": 0.0902,
+      "loss": 0.0452,
       "step": 9032
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.239632618981353e-06,
-      "loss": 0.2507,
+      "loss": 0.0511,
       "step": 9033
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.2377771592912146e-06,
-      "loss": 0.1405,
+      "loss": 0.0738,
       "step": 9034
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.235921699601076e-06,
-      "loss": 0.087,
+      "loss": 0.0896,
       "step": 9035
     },
     {
       "epoch": 2.51,
       "learning_rate": 3.234066239910938e-06,
-      "loss": 0.0322,
+      "loss": 0.0534,
       "step": 9036
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.2322107802207998e-06,
-      "loss": 0.193,
+      "loss": 0.0469,
       "step": 9037
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.2303553205306616e-06,
-      "loss": 0.1391,
+      "loss": 0.0749,
       "step": 9038
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.228499860840523e-06,
-      "loss": 0.138,
+      "loss": 0.0578,
       "step": 9039
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.226644401150385e-06,
-      "loss": 0.1416,
+      "loss": 0.0472,
       "step": 9040
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.224788941460247e-06,
-      "loss": 0.0847,
+      "loss": 0.0158,
       "step": 9041
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.2229334817701087e-06,
-      "loss": 0.085,
+      "loss": 0.0586,
       "step": 9042
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.22107802207997e-06,
-      "loss": 0.1385,
+      "loss": 0.0626,
       "step": 9043
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.219222562389832e-06,
-      "loss": 0.2461,
+      "loss": 0.0424,
       "step": 9044
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.217367102699694e-06,
-      "loss": 0.1913,
+      "loss": 0.0614,
       "step": 9045
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.2155116430095558e-06,
-      "loss": 0.0845,
+      "loss": 0.1045,
       "step": 9046
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.2136561833194176e-06,
-      "loss": 0.0319,
+      "loss": 0.1186,
       "step": 9047
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.211800723629279e-06,
-      "loss": 0.0307,
+      "loss": 0.1435,
       "step": 9048
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.209945263939141e-06,
-      "loss": 0.1394,
+      "loss": 0.033,
       "step": 9049
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.208089804249003e-06,
-      "loss": 0.2981,
+      "loss": 0.3016,
       "step": 9050
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.2062343445588647e-06,
-      "loss": 0.0311,
+      "loss": 0.1904,
       "step": 9051
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.204378884868726e-06,
-      "loss": 0.1959,
+      "loss": 0.0142,
       "step": 9052
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.202523425178588e-06,
-      "loss": 0.1351,
+      "loss": 0.1108,
       "step": 9053
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.20066796548845e-06,
-      "loss": 0.0853,
+      "loss": 0.0823,
       "step": 9054
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.1988125057983117e-06,
-      "loss": 0.2462,
+      "loss": 0.0763,
       "step": 9055
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.196957046108173e-06,
-      "loss": 0.1897,
+      "loss": 0.085,
       "step": 9056
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.195101586418035e-06,
-      "loss": 0.0857,
+      "loss": 0.0249,
       "step": 9057
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.193246126727897e-06,
-      "loss": 0.0867,
+      "loss": 0.0988,
       "step": 9058
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.191390667037759e-06,
-      "loss": 0.0844,
+      "loss": 0.0679,
       "step": 9059
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.1895352073476202e-06,
-      "loss": 0.0866,
+      "loss": 0.1058,
       "step": 9060
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.187679747657482e-06,
-      "loss": 0.1929,
+      "loss": 0.1264,
       "step": 9061
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.185824287967344e-06,
-      "loss": 0.1428,
+      "loss": 0.0669,
       "step": 9062
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.183968828277206e-06,
-      "loss": 0.0869,
+      "loss": 0.1435,
       "step": 9063
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.1821133685870677e-06,
-      "loss": 0.0837,
+      "loss": 0.0583,
       "step": 9064
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.180257908896929e-06,
-      "loss": 0.1951,
+      "loss": 0.0572,
       "step": 9065
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.178402449206791e-06,
-      "loss": 0.1978,
+      "loss": 0.1045,
       "step": 9066
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.176546989516653e-06,
-      "loss": 0.3027,
+      "loss": 0.1262,
       "step": 9067
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.1746915298265148e-06,
-      "loss": 0.0854,
+      "loss": 0.0552,
       "step": 9068
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.1728360701363762e-06,
-      "loss": 0.3013,
+      "loss": 0.0293,
       "step": 9069
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.170980610446238e-06,
-      "loss": 0.1403,
+      "loss": 0.0959,
       "step": 9070
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.1691251507561e-06,
-      "loss": 0.1917,
+      "loss": 0.06,
       "step": 9071
     },
     {
       "epoch": 2.52,
       "learning_rate": 3.167269691065962e-06,
-      "loss": 0.0881,
+      "loss": 0.0347,
       "step": 9072
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1654142313758233e-06,
-      "loss": 0.0844,
+      "loss": 0.1364,
       "step": 9073
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.163558771685685e-06,
-      "loss": 0.1934,
+      "loss": 0.0237,
       "step": 9074
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.161703311995547e-06,
-      "loss": 0.0304,
+      "loss": 0.0746,
       "step": 9075
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.159847852305409e-06,
-      "loss": 0.0861,
+      "loss": 0.0227,
       "step": 9076
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1579923926152708e-06,
-      "loss": 0.1954,
+      "loss": 0.0548,
       "step": 9077
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1561369329251322e-06,
-      "loss": 0.0862,
+      "loss": 0.0222,
       "step": 9078
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.154281473234994e-06,
-      "loss": 0.0833,
+      "loss": 0.0436,
       "step": 9079
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.152426013544856e-06,
-      "loss": 0.0846,
+      "loss": 0.0495,
       "step": 9080
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.150570553854718e-06,
-      "loss": 0.1929,
+      "loss": 0.045,
       "step": 9081
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1487150941645793e-06,
-      "loss": 0.0823,
+      "loss": 0.0164,
       "step": 9082
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.146859634474441e-06,
-      "loss": 0.2456,
+      "loss": 0.026,
       "step": 9083
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.145004174784303e-06,
-      "loss": 0.0854,
+      "loss": 0.0167,
       "step": 9084
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.143148715094165e-06,
-      "loss": 0.1367,
+      "loss": 0.0245,
       "step": 9085
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1412932554040263e-06,
-      "loss": 0.1387,
+      "loss": 0.0223,
       "step": 9086
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.139437795713888e-06,
-      "loss": 0.0876,
+      "loss": 0.1171,
       "step": 9087
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.13758233602375e-06,
-      "loss": 0.2505,
+      "loss": 0.0241,
       "step": 9088
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.135726876333612e-06,
-      "loss": 0.1386,
+      "loss": 0.0809,
       "step": 9089
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1338714166434734e-06,
-      "loss": 0.1898,
+      "loss": 0.0457,
       "step": 9090
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1320159569533353e-06,
-      "loss": 0.0302,
+      "loss": 0.0918,
       "step": 9091
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.130160497263197e-06,
-      "loss": 0.1385,
+      "loss": 0.0614,
       "step": 9092
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.128305037573059e-06,
-      "loss": 0.1399,
+      "loss": 0.0897,
       "step": 9093
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.126449577882921e-06,
-      "loss": 0.0305,
+      "loss": 0.0718,
       "step": 9094
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1245941181927823e-06,
-      "loss": 0.0855,
+      "loss": 0.0927,
       "step": 9095
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.122738658502644e-06,
-      "loss": 0.2487,
+      "loss": 0.1133,
       "step": 9096
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.120883198812506e-06,
-      "loss": 0.0817,
+      "loss": 0.0179,
       "step": 9097
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.119027739122368e-06,
-      "loss": 0.1401,
+      "loss": 0.0517,
       "step": 9098
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1171722794322294e-06,
-      "loss": 0.137,
+      "loss": 0.0826,
       "step": 9099
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1153168197420913e-06,
-      "loss": 0.0313,
+      "loss": 0.1236,
       "step": 9100
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.113461360051953e-06,
-      "loss": 0.1327,
+      "loss": 0.0596,
       "step": 9101
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.111605900361815e-06,
-      "loss": 0.1901,
+      "loss": 0.0182,
       "step": 9102
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1097504406716764e-06,
-      "loss": 0.138,
+      "loss": 0.0242,
       "step": 9103
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1078949809815383e-06,
-      "loss": 0.197,
+      "loss": 0.0646,
       "step": 9104
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1060395212914e-06,
-      "loss": 0.1367,
+      "loss": 0.0109,
       "step": 9105
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.104184061601262e-06,
-      "loss": 0.1405,
+      "loss": 0.1649,
       "step": 9106
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.102328601911124e-06,
-      "loss": 0.0843,
+      "loss": 0.0512,
       "step": 9107
     },
     {
       "epoch": 2.53,
       "learning_rate": 3.1004731422209854e-06,
-      "loss": 0.1924,
+      "loss": 0.0383,
       "step": 9108
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0986176825308472e-06,
-      "loss": 0.3548,
+      "loss": 0.0832,
       "step": 9109
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.096762222840709e-06,
-      "loss": 0.0309,
+      "loss": 0.0553,
       "step": 9110
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.094906763150571e-06,
-      "loss": 0.085,
+      "loss": 0.0199,
       "step": 9111
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0930513034604324e-06,
-      "loss": 0.0819,
+      "loss": 0.1449,
       "step": 9112
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0911958437702943e-06,
-      "loss": 0.0836,
+      "loss": 0.0109,
       "step": 9113
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.089340384080156e-06,
-      "loss": 0.0834,
+      "loss": 0.0697,
       "step": 9114
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.087484924390018e-06,
-      "loss": 0.1961,
+      "loss": 0.0692,
       "step": 9115
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0856294646998795e-06,
-      "loss": 0.136,
+      "loss": 0.1059,
       "step": 9116
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0837740050097414e-06,
-      "loss": 0.2473,
+      "loss": 0.047,
       "step": 9117
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0819185453196032e-06,
-      "loss": 0.0841,
+      "loss": 0.1268,
       "step": 9118
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.080063085629465e-06,
-      "loss": 0.0309,
+      "loss": 0.0143,
       "step": 9119
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0782076259393266e-06,
-      "loss": 0.0875,
+      "loss": 0.1336,
       "step": 9120
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0763521662491884e-06,
-      "loss": 0.1407,
+      "loss": 0.0099,
       "step": 9121
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0744967065590503e-06,
-      "loss": 0.1437,
+      "loss": 0.122,
       "step": 9122
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.072641246868912e-06,
-      "loss": 0.1961,
+      "loss": 0.0144,
       "step": 9123
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.070785787178774e-06,
-      "loss": 0.1415,
+      "loss": 0.0596,
       "step": 9124
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0689303274886355e-06,
-      "loss": 0.1366,
+      "loss": 0.0113,
       "step": 9125
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0670748677984974e-06,
-      "loss": 0.135,
+      "loss": 0.1015,
       "step": 9126
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0652194081083592e-06,
-      "loss": 0.1436,
+      "loss": 0.0155,
       "step": 9127
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.063363948418221e-06,
-      "loss": 0.0882,
+      "loss": 0.0873,
       "step": 9128
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0615084887280825e-06,
-      "loss": 0.1928,
+      "loss": 0.0646,
       "step": 9129
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0596530290379444e-06,
-      "loss": 0.0866,
+      "loss": 0.0159,
       "step": 9130
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0577975693478063e-06,
-      "loss": 0.1417,
+      "loss": 0.0163,
       "step": 9131
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.055942109657668e-06,
-      "loss": 0.0315,
+      "loss": 0.0382,
       "step": 9132
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0540866499675296e-06,
-      "loss": 0.1985,
+      "loss": 0.0237,
       "step": 9133
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0522311902773915e-06,
-      "loss": 0.1394,
+      "loss": 0.1258,
       "step": 9134
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0503757305872533e-06,
-      "loss": 0.2429,
+      "loss": 0.0413,
       "step": 9135
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.048520270897115e-06,
-      "loss": 0.1933,
+      "loss": 0.0387,
       "step": 9136
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.046664811206977e-06,
-      "loss": 0.1948,
+      "loss": 0.1631,
       "step": 9137
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0448093515168385e-06,
-      "loss": 0.2495,
+      "loss": 0.1916,
       "step": 9138
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0429538918267004e-06,
-      "loss": 0.0823,
+      "loss": 0.0737,
       "step": 9139
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0410984321365623e-06,
-      "loss": 0.1411,
+      "loss": 0.0239,
       "step": 9140
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.039242972446424e-06,
-      "loss": 0.1376,
+      "loss": 0.1031,
       "step": 9141
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0373875127562856e-06,
-      "loss": 0.1874,
+      "loss": 0.1975,
       "step": 9142
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0355320530661475e-06,
-      "loss": 0.0863,
+      "loss": 0.1117,
       "step": 9143
     },
     {
       "epoch": 2.54,
       "learning_rate": 3.0336765933760093e-06,
-      "loss": 0.0318,
+      "loss": 0.0582,
       "step": 9144
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.031821133685871e-06,
-      "loss": 0.2425,
+      "loss": 0.0172,
       "step": 9145
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0299656739957326e-06,
-      "loss": 0.1993,
+      "loss": 0.0247,
       "step": 9146
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0281102143055945e-06,
-      "loss": 0.0849,
+      "loss": 0.0955,
       "step": 9147
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0262547546154564e-06,
-      "loss": 0.1405,
+      "loss": 0.0245,
       "step": 9148
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0243992949253183e-06,
-      "loss": 0.0868,
+      "loss": 0.0388,
       "step": 9149
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0225438352351797e-06,
-      "loss": 0.1985,
+      "loss": 0.1758,
       "step": 9150
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0206883755450416e-06,
-      "loss": 0.2442,
+      "loss": 0.1367,
       "step": 9151
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0188329158549034e-06,
-      "loss": 0.1363,
+      "loss": 0.1143,
       "step": 9152
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0169774561647653e-06,
-      "loss": 0.032,
+      "loss": 0.0189,
       "step": 9153
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.015121996474627e-06,
-      "loss": 0.0819,
+      "loss": 0.1093,
       "step": 9154
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0132665367844886e-06,
-      "loss": 0.1392,
+      "loss": 0.0203,
       "step": 9155
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0114110770943505e-06,
-      "loss": 0.086,
+      "loss": 0.126,
       "step": 9156
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0095556174042124e-06,
-      "loss": 0.1393,
+      "loss": 0.0903,
       "step": 9157
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0077001577140742e-06,
-      "loss": 0.0324,
+      "loss": 0.1127,
       "step": 9158
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0058446980239357e-06,
-      "loss": 0.1357,
+      "loss": 0.0523,
       "step": 9159
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0039892383337976e-06,
-      "loss": 0.1422,
+      "loss": 0.0208,
       "step": 9160
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0021337786436594e-06,
-      "loss": 0.0313,
+      "loss": 0.0528,
       "step": 9161
     },
     {
       "epoch": 2.55,
       "learning_rate": 3.0002783189535213e-06,
-      "loss": 0.4174,
+      "loss": 0.0121,
       "step": 9162
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9984228592633828e-06,
-      "loss": 0.192,
+      "loss": 0.1026,
       "step": 9163
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9965673995732446e-06,
-      "loss": 0.0848,
+      "loss": 0.0687,
       "step": 9164
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9947119398831065e-06,
-      "loss": 0.1956,
+      "loss": 0.017,
       "step": 9165
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9928564801929684e-06,
-      "loss": 0.2986,
+      "loss": 0.1171,
       "step": 9166
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9910010205028302e-06,
-      "loss": 0.2461,
+      "loss": 0.0381,
       "step": 9167
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9891455608126917e-06,
-      "loss": 0.139,
+      "loss": 0.0207,
       "step": 9168
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9872901011225536e-06,
-      "loss": 0.1358,
+      "loss": 0.2333,
       "step": 9169
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9854346414324154e-06,
-      "loss": 0.2493,
+      "loss": 0.0231,
       "step": 9170
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9835791817422773e-06,
-      "loss": 0.245,
+      "loss": 0.0136,
       "step": 9171
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9817237220521387e-06,
-      "loss": 0.0322,
+      "loss": 0.1316,
       "step": 9172
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9798682623620006e-06,
-      "loss": 0.246,
+      "loss": 0.1129,
       "step": 9173
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9780128026718625e-06,
-      "loss": 0.0326,
+      "loss": 0.0522,
       "step": 9174
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9761573429817244e-06,
-      "loss": 0.1396,
+      "loss": 0.0412,
       "step": 9175
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.974301883291586e-06,
-      "loss": 0.088,
+      "loss": 0.0602,
       "step": 9176
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9724464236014477e-06,
-      "loss": 0.1891,
+      "loss": 0.0687,
       "step": 9177
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9705909639113095e-06,
-      "loss": 0.1949,
+      "loss": 0.0382,
       "step": 9178
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.9687355042211714e-06,
-      "loss": 0.0338,
+      "loss": 0.0327,
       "step": 9179
     },
     {
       "epoch": 2.55,
       "learning_rate": 2.966880044531033e-06,
-      "loss": 0.1935,
+      "loss": 0.127,
       "step": 9180
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9650245848408947e-06,
-      "loss": 0.1891,
+      "loss": 0.0461,
       "step": 9181
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9631691251507566e-06,
-      "loss": 0.3034,
+      "loss": 0.0861,
       "step": 9182
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9613136654606185e-06,
-      "loss": 0.0861,
+      "loss": 0.0164,
       "step": 9183
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9594582057704803e-06,
-      "loss": 0.1351,
+      "loss": 0.0147,
       "step": 9184
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9576027460803414e-06,
-      "loss": 0.1378,
+      "loss": 0.015,
       "step": 9185
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9557472863902032e-06,
-      "loss": 0.0346,
+      "loss": 0.0164,
       "step": 9186
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.953891826700065e-06,
-      "loss": 0.1393,
+      "loss": 0.1492,
       "step": 9187
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9520363670099266e-06,
-      "loss": 0.1374,
+      "loss": 0.1612,
       "step": 9188
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9501809073197884e-06,
-      "loss": 0.0877,
+      "loss": 0.1191,
       "step": 9189
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9483254476296503e-06,
-      "loss": 0.0894,
+      "loss": 0.0468,
       "step": 9190
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.946469987939512e-06,
-      "loss": 0.1429,
+      "loss": 0.0442,
       "step": 9191
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9446145282493736e-06,
-      "loss": 0.0844,
+      "loss": 0.0527,
       "step": 9192
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9427590685592355e-06,
-      "loss": 0.0883,
+      "loss": 0.1261,
       "step": 9193
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9409036088690974e-06,
-      "loss": 0.034,
+      "loss": 0.0198,
       "step": 9194
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9390481491789592e-06,
-      "loss": 0.0847,
+      "loss": 0.0191,
       "step": 9195
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9371926894888207e-06,
-      "loss": 0.084,
+      "loss": 0.1045,
       "step": 9196
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9353372297986825e-06,
-      "loss": 0.0333,
+      "loss": 0.0428,
       "step": 9197
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9334817701085444e-06,
-      "loss": 0.0882,
+      "loss": 0.0155,
       "step": 9198
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9316263104184063e-06,
-      "loss": 0.0881,
+      "loss": 0.0157,
       "step": 9199
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.929770850728268e-06,
-      "loss": 0.0833,
+      "loss": 0.1016,
       "step": 9200
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9279153910381296e-06,
-      "loss": 0.1415,
+      "loss": 0.0492,
       "step": 9201
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9260599313479915e-06,
-      "loss": 0.1406,
+      "loss": 0.0198,
       "step": 9202
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9242044716578533e-06,
-      "loss": 0.0845,
+      "loss": 0.0894,
       "step": 9203
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.922349011967715e-06,
-      "loss": 0.1968,
+      "loss": 0.1425,
       "step": 9204
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9204935522775767e-06,
-      "loss": 0.0314,
+      "loss": 0.0584,
       "step": 9205
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9186380925874385e-06,
-      "loss": 0.1926,
+      "loss": 0.0511,
       "step": 9206
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9167826328973004e-06,
-      "loss": 0.2497,
+      "loss": 0.022,
       "step": 9207
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9149271732071623e-06,
-      "loss": 0.0307,
+      "loss": 0.0448,
       "step": 9208
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9130717135170237e-06,
-      "loss": 0.1958,
+      "loss": 0.018,
       "step": 9209
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9112162538268856e-06,
-      "loss": 0.2002,
+      "loss": 0.0294,
       "step": 9210
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9093607941367475e-06,
-      "loss": 0.1362,
+      "loss": 0.0949,
       "step": 9211
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9075053344466093e-06,
-      "loss": 0.2433,
+      "loss": 0.0355,
       "step": 9212
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9056498747564708e-06,
-      "loss": 0.0856,
+      "loss": 0.0547,
       "step": 9213
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9037944150663326e-06,
-      "loss": 0.0851,
+      "loss": 0.062,
       "step": 9214
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9019389553761945e-06,
-      "loss": 0.1955,
+      "loss": 0.0943,
       "step": 9215
     },
     {
       "epoch": 2.56,
       "learning_rate": 2.9000834956860564e-06,
-      "loss": 0.0821,
+      "loss": 0.0605,
       "step": 9216
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8982280359959183e-06,
-      "loss": 0.0876,
+      "loss": 0.0184,
       "step": 9217
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8963725763057797e-06,
-      "loss": 0.138,
+      "loss": 0.0146,
       "step": 9218
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8945171166156416e-06,
-      "loss": 0.0295,
+      "loss": 0.094,
       "step": 9219
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8926616569255034e-06,
-      "loss": 0.0855,
+      "loss": 0.0157,
       "step": 9220
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8908061972353653e-06,
-      "loss": 0.0854,
+      "loss": 0.015,
       "step": 9221
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8889507375452268e-06,
-      "loss": 0.1414,
+      "loss": 0.0123,
       "step": 9222
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8870952778550886e-06,
-      "loss": 0.1373,
+      "loss": 0.0884,
       "step": 9223
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8852398181649505e-06,
-      "loss": 0.1388,
+      "loss": 0.0465,
       "step": 9224
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8833843584748124e-06,
-      "loss": 0.1387,
+      "loss": 0.0946,
       "step": 9225
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.881528898784674e-06,
-      "loss": 0.1375,
+      "loss": 0.0075,
       "step": 9226
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8796734390945357e-06,
-      "loss": 0.14,
+      "loss": 0.2284,
       "step": 9227
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8778179794043976e-06,
-      "loss": 0.1393,
+      "loss": 0.0973,
       "step": 9228
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8759625197142594e-06,
-      "loss": 0.0308,
+      "loss": 0.01,
       "step": 9229
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8741070600241213e-06,
-      "loss": 0.1902,
+      "loss": 0.1158,
       "step": 9230
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8722516003339827e-06,
-      "loss": 0.1934,
+      "loss": 0.0703,
       "step": 9231
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8703961406438446e-06,
-      "loss": 0.1989,
+      "loss": 0.0187,
       "step": 9232
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8685406809537065e-06,
-      "loss": 0.086,
+      "loss": 0.0201,
       "step": 9233
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8666852212635684e-06,
-      "loss": 0.1363,
+      "loss": 0.0773,
       "step": 9234
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.86482976157343e-06,
-      "loss": 0.0833,
+      "loss": 0.0395,
       "step": 9235
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8629743018832917e-06,
-      "loss": 0.1425,
+      "loss": 0.0647,
       "step": 9236
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8611188421931535e-06,
-      "loss": 0.0298,
+      "loss": 0.0121,
       "step": 9237
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8592633825030154e-06,
-      "loss": 0.1407,
+      "loss": 0.0085,
       "step": 9238
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.857407922812877e-06,
-      "loss": 0.0294,
+      "loss": 0.0589,
       "step": 9239
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8555524631227387e-06,
-      "loss": 0.0833,
+      "loss": 0.0824,
       "step": 9240
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8536970034326006e-06,
-      "loss": 0.084,
+      "loss": 0.0711,
       "step": 9241
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8518415437424625e-06,
-      "loss": 0.0847,
+      "loss": 0.0603,
       "step": 9242
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.849986084052324e-06,
-      "loss": 0.0288,
+      "loss": 0.0882,
       "step": 9243
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.848130624362186e-06,
-      "loss": 0.2532,
+      "loss": 0.2251,
       "step": 9244
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8462751646720477e-06,
-      "loss": 0.085,
+      "loss": 0.0162,
       "step": 9245
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8444197049819095e-06,
-      "loss": 0.1329,
+      "loss": 0.0486,
       "step": 9246
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8425642452917714e-06,
-      "loss": 0.2515,
+      "loss": 0.0585,
       "step": 9247
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.840708785601633e-06,
-      "loss": 0.2571,
+      "loss": 0.015,
       "step": 9248
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8388533259114947e-06,
-      "loss": 0.1396,
+      "loss": 0.0354,
       "step": 9249
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8369978662213566e-06,
-      "loss": 0.3622,
+      "loss": 0.0802,
       "step": 9250
     },
     {
       "epoch": 2.57,
       "learning_rate": 2.8351424065312185e-06,
-      "loss": 0.0852,
+      "loss": 0.0122,
       "step": 9251
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.83328694684108e-06,
-      "loss": 0.0286,
+      "loss": 0.0583,
       "step": 9252
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.8314314871509418e-06,
-      "loss": 0.1981,
+      "loss": 0.0643,
       "step": 9253
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.8295760274608037e-06,
-      "loss": 0.2473,
+      "loss": 0.0535,
       "step": 9254
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.8277205677706655e-06,
-      "loss": 0.0298,
+      "loss": 0.0629,
       "step": 9255
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.825865108080527e-06,
-      "loss": 0.1397,
+      "loss": 0.0273,
       "step": 9256
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.824009648390389e-06,
-      "loss": 0.0829,
+      "loss": 0.0839,
       "step": 9257
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.8221541887002507e-06,
-      "loss": 0.1389,
+      "loss": 0.0491,
       "step": 9258
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.8202987290101126e-06,
-      "loss": 0.0831,
+      "loss": 0.0521,
       "step": 9259
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.8184432693199745e-06,
-      "loss": 0.1385,
+      "loss": 0.1103,
       "step": 9260
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.816587809629836e-06,
-      "loss": 0.1432,
+      "loss": 0.1207,
       "step": 9261
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.8147323499396978e-06,
-      "loss": 0.0294,
+      "loss": 0.0115,
       "step": 9262
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.8128768902495596e-06,
-      "loss": 0.1389,
+      "loss": 0.1834,
       "step": 9263
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.8110214305594215e-06,
-      "loss": 0.251,
+      "loss": 0.0569,
       "step": 9264
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.809165970869283e-06,
-      "loss": 0.086,
+      "loss": 0.0136,
       "step": 9265
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.807310511179145e-06,
-      "loss": 0.1905,
+      "loss": 0.0814,
       "step": 9266
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.8054550514890067e-06,
-      "loss": 0.1964,
+      "loss": 0.0171,
       "step": 9267
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.8035995917988686e-06,
-      "loss": 0.1385,
+      "loss": 0.0476,
       "step": 9268
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.80174413210873e-06,
-      "loss": 0.0836,
+      "loss": 0.0192,
       "step": 9269
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.799888672418592e-06,
-      "loss": 0.2523,
+      "loss": 0.0122,
       "step": 9270
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.7980332127284538e-06,
-      "loss": 0.0815,
+      "loss": 0.1772,
       "step": 9271
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.7961777530383156e-06,
-      "loss": 0.1933,
+      "loss": 0.0396,
       "step": 9272
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.794322293348177e-06,
-      "loss": 0.1376,
+      "loss": 0.0877,
       "step": 9273
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.792466833658039e-06,
-      "loss": 0.0862,
+      "loss": 0.0144,
       "step": 9274
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.790611373967901e-06,
-      "loss": 0.3041,
+      "loss": 0.0369,
       "step": 9275
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.7887559142777627e-06,
-      "loss": 0.0836,
+      "loss": 0.0972,
       "step": 9276
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.7869004545876246e-06,
-      "loss": 0.1375,
+      "loss": 0.0612,
       "step": 9277
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.785044994897486e-06,
-      "loss": 0.1974,
+      "loss": 0.0466,
       "step": 9278
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.783189535207348e-06,
-      "loss": 0.0301,
+      "loss": 0.0512,
       "step": 9279
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.7813340755172097e-06,
-      "loss": 0.197,
+      "loss": 0.0755,
       "step": 9280
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.7794786158270716e-06,
-      "loss": 0.1944,
+      "loss": 0.0982,
       "step": 9281
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.777623156136933e-06,
-      "loss": 0.1932,
+      "loss": 0.0702,
       "step": 9282
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.775767696446795e-06,
-      "loss": 0.1388,
+      "loss": 0.0686,
       "step": 9283
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.773912236756657e-06,
-      "loss": 0.0301,
+      "loss": 0.0455,
       "step": 9284
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.7720567770665187e-06,
-      "loss": 0.1378,
+      "loss": 0.0419,
       "step": 9285
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.77020131737638e-06,
-      "loss": 0.0301,
+      "loss": 0.0921,
       "step": 9286
     },
     {
       "epoch": 2.58,
       "learning_rate": 2.768345857686242e-06,
-      "loss": 0.2518,
+      "loss": 0.0289,
       "step": 9287
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.766490397996104e-06,
-      "loss": 0.1985,
+      "loss": 0.0823,
       "step": 9288
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.7646349383059657e-06,
-      "loss": 0.1381,
+      "loss": 0.102,
       "step": 9289
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.7627794786158276e-06,
-      "loss": 0.3005,
+      "loss": 0.0628,
       "step": 9290
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.760924018925689e-06,
-      "loss": 0.0314,
+      "loss": 0.0125,
       "step": 9291
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.759068559235551e-06,
-      "loss": 0.1382,
+      "loss": 0.0201,
       "step": 9292
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.757213099545413e-06,
-      "loss": 0.141,
+      "loss": 0.0982,
       "step": 9293
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.7553576398552747e-06,
-      "loss": 0.251,
+      "loss": 0.114,
       "step": 9294
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.753502180165136e-06,
-      "loss": 0.1354,
+      "loss": 0.122,
       "step": 9295
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.751646720474998e-06,
-      "loss": 0.0317,
+      "loss": 0.0595,
       "step": 9296
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.74979126078486e-06,
-      "loss": 0.0312,
+      "loss": 0.1707,
       "step": 9297
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.7479358010947217e-06,
-      "loss": 0.2434,
+      "loss": 0.0651,
       "step": 9298
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.746080341404583e-06,
-      "loss": 0.0847,
+      "loss": 0.0157,
       "step": 9299
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.744224881714445e-06,
-      "loss": 0.3557,
+      "loss": 0.0832,
       "step": 9300
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.742369422024307e-06,
-      "loss": 0.2468,
+      "loss": 0.0203,
       "step": 9301
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.7405139623341688e-06,
-      "loss": 0.1948,
+      "loss": 0.1308,
       "step": 9302
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.7386585026440302e-06,
-      "loss": 0.0866,
+      "loss": 0.0553,
       "step": 9303
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.736803042953892e-06,
-      "loss": 0.0312,
+      "loss": 0.1168,
       "step": 9304
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.734947583263754e-06,
-      "loss": 0.1382,
+      "loss": 0.0601,
       "step": 9305
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.733092123573616e-06,
-      "loss": 0.0865,
+      "loss": 0.0174,
       "step": 9306
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.7312366638834777e-06,
-      "loss": 0.1909,
+      "loss": 0.0904,
       "step": 9307
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.729381204193339e-06,
-      "loss": 0.0859,
+      "loss": 0.0164,
       "step": 9308
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.727525744503201e-06,
-      "loss": 0.2483,
+      "loss": 0.1479,
       "step": 9309
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.725670284813063e-06,
-      "loss": 0.0326,
+      "loss": 0.0148,
       "step": 9310
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.7238148251229248e-06,
-      "loss": 0.1407,
+      "loss": 0.0175,
       "step": 9311
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.7219593654327862e-06,
-      "loss": 0.0853,
+      "loss": 0.0807,
       "step": 9312
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.720103905742648e-06,
-      "loss": 0.0872,
+      "loss": 0.0471,
       "step": 9313
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.71824844605251e-06,
-      "loss": 0.245,
+      "loss": 0.0252,
       "step": 9314
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.716392986362372e-06,
-      "loss": 0.0837,
+      "loss": 0.0581,
       "step": 9315
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.7145375266722333e-06,
-      "loss": 0.084,
+      "loss": 0.0103,
       "step": 9316
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.712682066982095e-06,
-      "loss": 0.1406,
+      "loss": 0.0904,
       "step": 9317
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.710826607291957e-06,
-      "loss": 0.2484,
+      "loss": 0.0707,
       "step": 9318
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.708971147601819e-06,
-      "loss": 0.0878,
+      "loss": 0.0686,
       "step": 9319
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.7071156879116808e-06,
-      "loss": 0.2478,
+      "loss": 0.0653,
       "step": 9320
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.705260228221542e-06,
-      "loss": 0.1398,
+      "loss": 0.1608,
       "step": 9321
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.703404768531404e-06,
-      "loss": 0.2971,
+      "loss": 0.0191,
       "step": 9322
     },
     {
       "epoch": 2.59,
       "learning_rate": 2.701549308841266e-06,
-      "loss": 0.0845,
+      "loss": 0.0543,
       "step": 9323
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.699693849151128e-06,
-      "loss": 0.1409,
+      "loss": 0.0576,
       "step": 9324
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.6978383894609893e-06,
-      "loss": 0.1406,
+      "loss": 0.0577,
       "step": 9325
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.695982929770851e-06,
-      "loss": 0.0861,
+      "loss": 0.0943,
       "step": 9326
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.694127470080713e-06,
-      "loss": 0.0844,
+      "loss": 0.02,
       "step": 9327
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.692272010390575e-06,
-      "loss": 0.1383,
+      "loss": 0.0143,
       "step": 9328
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.6904165507004363e-06,
-      "loss": 0.14,
+      "loss": 0.0198,
       "step": 9329
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.688561091010298e-06,
-      "loss": 0.1396,
+      "loss": 0.0112,
       "step": 9330
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.68670563132016e-06,
-      "loss": 0.1377,
+      "loss": 0.2465,
       "step": 9331
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.684850171630022e-06,
-      "loss": 0.1412,
+      "loss": 0.271,
       "step": 9332
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.6829947119398834e-06,
-      "loss": 0.1369,
+      "loss": 0.0424,
       "step": 9333
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.6811392522497453e-06,
-      "loss": 0.197,
+      "loss": 0.0746,
       "step": 9334
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.679283792559607e-06,
-      "loss": 0.032,
+      "loss": 0.085,
       "step": 9335
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.677428332869469e-06,
-      "loss": 0.2461,
+      "loss": 0.0449,
       "step": 9336
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.67557287317933e-06,
-      "loss": 0.1379,
+      "loss": 0.0483,
       "step": 9337
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.673717413489192e-06,
-      "loss": 0.1352,
+      "loss": 0.1389,
       "step": 9338
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.6718619537990538e-06,
-      "loss": 0.0851,
+      "loss": 0.0441,
       "step": 9339
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.6700064941089156e-06,
-      "loss": 0.2964,
+      "loss": 0.0872,
       "step": 9340
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.668151034418777e-06,
-      "loss": 0.1387,
+      "loss": 0.0433,
       "step": 9341
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.666295574728639e-06,
-      "loss": 0.1378,
+      "loss": 0.0222,
       "step": 9342
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.664440115038501e-06,
-      "loss": 0.1375,
+      "loss": 0.1809,
       "step": 9343
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.6625846553483627e-06,
-      "loss": 0.1437,
+      "loss": 0.0983,
       "step": 9344
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.660729195658224e-06,
-      "loss": 0.0882,
+      "loss": 0.0252,
       "step": 9345
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.658873735968086e-06,
-      "loss": 0.2982,
+      "loss": 0.0602,
       "step": 9346
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.657018276277948e-06,
-      "loss": 0.087,
+      "loss": 0.0718,
       "step": 9347
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.6551628165878097e-06,
-      "loss": 0.1368,
+      "loss": 0.1584,
       "step": 9348
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.653307356897671e-06,
-      "loss": 0.1433,
+      "loss": 0.1003,
       "step": 9349
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.651451897207533e-06,
-      "loss": 0.0887,
+      "loss": 0.0786,
       "step": 9350
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.649596437517395e-06,
-      "loss": 0.3014,
+      "loss": 0.1346,
       "step": 9351
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.647740977827257e-06,
-      "loss": 0.0858,
+      "loss": 0.2276,
       "step": 9352
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.6458855181371187e-06,
-      "loss": 0.1973,
+      "loss": 0.0627,
       "step": 9353
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.64403005844698e-06,
-      "loss": 0.2408,
+      "loss": 0.0219,
       "step": 9354
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.642174598756842e-06,
-      "loss": 0.1407,
+      "loss": 0.102,
       "step": 9355
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.640319139066704e-06,
-      "loss": 0.1398,
+      "loss": 0.027,
       "step": 9356
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.6384636793765657e-06,
-      "loss": 0.0335,
+      "loss": 0.1838,
       "step": 9357
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.636608219686427e-06,
-      "loss": 0.238,
+      "loss": 0.029,
       "step": 9358
     },
     {
       "epoch": 2.6,
       "learning_rate": 2.634752759996289e-06,
-      "loss": 0.1919,
+      "loss": 0.02,
       "step": 9359
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.632897300306151e-06,
-      "loss": 0.0326,
+      "loss": 0.0804,
       "step": 9360
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.631041840616013e-06,
-      "loss": 0.2413,
+      "loss": 0.0176,
       "step": 9361
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.6291863809258742e-06,
-      "loss": 0.1418,
+      "loss": 0.0671,
       "step": 9362
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.627330921235736e-06,
-      "loss": 0.1911,
+      "loss": 0.0571,
       "step": 9363
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.625475461545598e-06,
-      "loss": 0.1377,
+      "loss": 0.0092,
       "step": 9364
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.62362000185546e-06,
-      "loss": 0.1436,
+      "loss": 0.1102,
       "step": 9365
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.6217645421653213e-06,
-      "loss": 0.3509,
+      "loss": 0.0686,
       "step": 9366
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.619909082475183e-06,
-      "loss": 0.1417,
+      "loss": 0.0336,
       "step": 9367
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.618053622785045e-06,
-      "loss": 0.1395,
+      "loss": 0.0859,
       "step": 9368
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.616198163094907e-06,
-      "loss": 0.1909,
+      "loss": 0.1805,
       "step": 9369
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.6143427034047688e-06,
-      "loss": 0.198,
+      "loss": 0.0411,
       "step": 9370
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.6124872437146302e-06,
-      "loss": 0.1396,
+      "loss": 0.0965,
       "step": 9371
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.610631784024492e-06,
-      "loss": 0.1373,
+      "loss": 0.0866,
       "step": 9372
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.608776324334354e-06,
-      "loss": 0.1918,
+      "loss": 0.0601,
       "step": 9373
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.606920864644216e-06,
-      "loss": 0.0887,
+      "loss": 0.103,
       "step": 9374
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.6050654049540773e-06,
-      "loss": 0.0875,
+      "loss": 0.1098,
       "step": 9375
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.603209945263939e-06,
-      "loss": 0.1926,
+      "loss": 0.0603,
       "step": 9376
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.601354485573801e-06,
-      "loss": 0.0347,
+      "loss": 0.1624,
       "step": 9377
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.599499025883663e-06,
-      "loss": 0.1921,
+      "loss": 0.0632,
       "step": 9378
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.5976435661935243e-06,
-      "loss": 0.2999,
+      "loss": 0.022,
       "step": 9379
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.5957881065033862e-06,
-      "loss": 0.0849,
+      "loss": 0.0848,
       "step": 9380
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.593932646813248e-06,
-      "loss": 0.1388,
+      "loss": 0.0186,
       "step": 9381
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.59207718712311e-06,
-      "loss": 0.1921,
+      "loss": 0.059,
       "step": 9382
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.590221727432972e-06,
-      "loss": 0.1421,
+      "loss": 0.113,
       "step": 9383
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.5883662677428333e-06,
-      "loss": 0.0344,
+      "loss": 0.0883,
       "step": 9384
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.586510808052695e-06,
-      "loss": 0.1403,
+      "loss": 0.204,
       "step": 9385
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.584655348362557e-06,
-      "loss": 0.1952,
+      "loss": 0.096,
       "step": 9386
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.582799888672419e-06,
-      "loss": 0.2403,
+      "loss": 0.0906,
       "step": 9387
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.5809444289822803e-06,
-      "loss": 0.0884,
+      "loss": 0.0219,
       "step": 9388
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.579088969292142e-06,
-      "loss": 0.1398,
+      "loss": 0.0987,
       "step": 9389
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.577233509602004e-06,
-      "loss": 0.1373,
+      "loss": 0.1263,
       "step": 9390
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.575378049911866e-06,
-      "loss": 0.141,
+      "loss": 0.1234,
       "step": 9391
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.5735225902217274e-06,
-      "loss": 0.0846,
+      "loss": 0.0552,
       "step": 9392
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.5716671305315893e-06,
-      "loss": 0.0873,
+      "loss": 0.0159,
       "step": 9393
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.569811670841451e-06,
-      "loss": 0.0343,
+      "loss": 0.0219,
       "step": 9394
     },
     {
       "epoch": 2.61,
       "learning_rate": 2.567956211151313e-06,
-      "loss": 0.0847,
+      "loss": 0.1053,
       "step": 9395
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5661007514611745e-06,
-      "loss": 0.0341,
+      "loss": 0.0213,
       "step": 9396
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5642452917710363e-06,
-      "loss": 0.1372,
+      "loss": 0.1149,
       "step": 9397
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.562389832080898e-06,
-      "loss": 0.2961,
+      "loss": 0.0357,
       "step": 9398
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.56053437239076e-06,
-      "loss": 0.0347,
+      "loss": 0.0907,
       "step": 9399
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.558678912700622e-06,
-      "loss": 0.0877,
+      "loss": 0.0096,
       "step": 9400
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5568234530104834e-06,
-      "loss": 0.1413,
+      "loss": 0.1576,
       "step": 9401
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5549679933203453e-06,
-      "loss": 0.1413,
+      "loss": 0.0225,
       "step": 9402
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.553112533630207e-06,
-      "loss": 0.0338,
+      "loss": 0.0689,
       "step": 9403
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.551257073940069e-06,
-      "loss": 0.0874,
+      "loss": 0.0787,
       "step": 9404
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5494016142499304e-06,
-      "loss": 0.199,
+      "loss": 0.0471,
       "step": 9405
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5475461545597923e-06,
-      "loss": 0.1403,
+      "loss": 0.0137,
       "step": 9406
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.545690694869654e-06,
-      "loss": 0.137,
+      "loss": 0.084,
       "step": 9407
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.543835235179516e-06,
-      "loss": 0.1881,
+      "loss": 0.2517,
       "step": 9408
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5419797754893775e-06,
-      "loss": 0.2465,
+      "loss": 0.076,
       "step": 9409
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5401243157992394e-06,
-      "loss": 0.0885,
+      "loss": 0.0118,
       "step": 9410
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5382688561091012e-06,
-      "loss": 0.0338,
+      "loss": 0.0336,
       "step": 9411
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.536413396418963e-06,
-      "loss": 0.088,
+      "loss": 0.1218,
       "step": 9412
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.534557936728825e-06,
-      "loss": 0.1887,
+      "loss": 0.0688,
       "step": 9413
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5327024770386864e-06,
-      "loss": 0.0873,
+      "loss": 0.0362,
       "step": 9414
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5308470173485483e-06,
-      "loss": 0.1391,
+      "loss": 0.053,
       "step": 9415
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.52899155765841e-06,
-      "loss": 0.0852,
+      "loss": 0.0679,
       "step": 9416
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.527136097968272e-06,
-      "loss": 0.1382,
+      "loss": 0.0126,
       "step": 9417
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5252806382781335e-06,
-      "loss": 0.1922,
+      "loss": 0.0242,
       "step": 9418
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5234251785879954e-06,
-      "loss": 0.0327,
+      "loss": 0.0111,
       "step": 9419
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5215697188978572e-06,
-      "loss": 0.1414,
+      "loss": 0.1102,
       "step": 9420
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.519714259207719e-06,
-      "loss": 0.0318,
+      "loss": 0.0514,
       "step": 9421
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5178587995175805e-06,
-      "loss": 0.0891,
+      "loss": 0.0208,
       "step": 9422
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5160033398274424e-06,
-      "loss": 0.1345,
+      "loss": 0.0162,
       "step": 9423
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5141478801373043e-06,
-      "loss": 0.0314,
+      "loss": 0.0189,
       "step": 9424
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.512292420447166e-06,
-      "loss": 0.1369,
+      "loss": 0.094,
       "step": 9425
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5104369607570276e-06,
-      "loss": 0.2391,
+      "loss": 0.2002,
       "step": 9426
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5085815010668895e-06,
-      "loss": 0.1427,
+      "loss": 0.0145,
       "step": 9427
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5067260413767513e-06,
-      "loss": 0.0849,
+      "loss": 0.0086,
       "step": 9428
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5048705816866132e-06,
-      "loss": 0.0311,
+      "loss": 0.0451,
       "step": 9429
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.503015121996475e-06,
-      "loss": 0.3087,
+      "loss": 0.0586,
       "step": 9430
     },
     {
       "epoch": 2.62,
       "learning_rate": 2.5011596623063365e-06,
-      "loss": 0.0858,
+      "loss": 0.0593,
       "step": 9431
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4993042026161984e-06,
-      "loss": 0.0873,
+      "loss": 0.0259,
       "step": 9432
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4974487429260603e-06,
-      "loss": 0.0851,
+      "loss": 0.0586,
       "step": 9433
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.495593283235922e-06,
-      "loss": 0.0863,
+      "loss": 0.0169,
       "step": 9434
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4937378235457836e-06,
-      "loss": 0.1969,
+      "loss": 0.0644,
       "step": 9435
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4918823638556455e-06,
-      "loss": 0.3034,
+      "loss": 0.0318,
       "step": 9436
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4900269041655073e-06,
-      "loss": 0.1933,
+      "loss": 0.0373,
       "step": 9437
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.488171444475369e-06,
-      "loss": 0.0836,
+      "loss": 0.0292,
       "step": 9438
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4863159847852307e-06,
-      "loss": 0.1383,
+      "loss": 0.0688,
       "step": 9439
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4844605250950925e-06,
-      "loss": 0.0831,
+      "loss": 0.0207,
       "step": 9440
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4826050654049544e-06,
-      "loss": 0.1384,
+      "loss": 0.1256,
       "step": 9441
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4807496057148163e-06,
-      "loss": 0.0314,
+      "loss": 0.1065,
       "step": 9442
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.478894146024678e-06,
-      "loss": 0.0304,
+      "loss": 0.0159,
       "step": 9443
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4770386863345396e-06,
-      "loss": 0.0851,
+      "loss": 0.058,
       "step": 9444
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4751832266444015e-06,
-      "loss": 0.139,
+      "loss": 0.0114,
       "step": 9445
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4733277669542633e-06,
-      "loss": 0.0822,
+      "loss": 0.0736,
       "step": 9446
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.471472307264125e-06,
-      "loss": 0.137,
+      "loss": 0.0529,
       "step": 9447
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4696168475739866e-06,
-      "loss": 0.1953,
+      "loss": 0.0112,
       "step": 9448
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4677613878838485e-06,
-      "loss": 0.1444,
+      "loss": 0.0146,
       "step": 9449
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4659059281937104e-06,
-      "loss": 0.1366,
+      "loss": 0.1063,
       "step": 9450
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.464050468503572e-06,
-      "loss": 0.0838,
+      "loss": 0.0878,
       "step": 9451
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4621950088134337e-06,
-      "loss": 0.0839,
+      "loss": 0.0629,
       "step": 9452
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4603395491232956e-06,
-      "loss": 0.0843,
+      "loss": 0.0284,
       "step": 9453
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.458484089433157e-06,
-      "loss": 0.1412,
+      "loss": 0.1068,
       "step": 9454
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.456628629743019e-06,
-      "loss": 0.0308,
+      "loss": 0.0684,
       "step": 9455
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4547731700528808e-06,
-      "loss": 0.0825,
+      "loss": 0.0117,
       "step": 9456
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4529177103627426e-06,
-      "loss": 0.1918,
+      "loss": 0.0715,
       "step": 9457
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.451062250672604e-06,
-      "loss": 0.03,
+      "loss": 0.0603,
       "step": 9458
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.449206790982466e-06,
-      "loss": 0.0844,
+      "loss": 0.0948,
       "step": 9459
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.447351331292328e-06,
-      "loss": 0.3039,
+      "loss": 0.0953,
       "step": 9460
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4454958716021897e-06,
-      "loss": 0.1971,
+      "loss": 0.0156,
       "step": 9461
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.443640411912051e-06,
-      "loss": 0.1392,
+      "loss": 0.0915,
       "step": 9462
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.441784952221913e-06,
-      "loss": 0.0874,
+      "loss": 0.0518,
       "step": 9463
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.439929492531775e-06,
-      "loss": 0.0301,
+      "loss": 0.0537,
       "step": 9464
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.4380740328416367e-06,
-      "loss": 0.0862,
+      "loss": 0.0095,
       "step": 9465
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.436218573151498e-06,
-      "loss": 0.0867,
+      "loss": 0.0195,
       "step": 9466
     },
     {
       "epoch": 2.63,
       "learning_rate": 2.43436311346136e-06,
-      "loss": 0.0867,
+      "loss": 0.1344,
       "step": 9467
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.432507653771222e-06,
-      "loss": 0.1919,
+      "loss": 0.0335,
       "step": 9468
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.430652194081084e-06,
-      "loss": 0.19,
+      "loss": 0.0541,
       "step": 9469
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.4287967343909457e-06,
-      "loss": 0.1401,
+      "loss": 0.0928,
       "step": 9470
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.426941274700807e-06,
-      "loss": 0.0844,
+      "loss": 0.0585,
       "step": 9471
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.425085815010669e-06,
-      "loss": 0.1368,
+      "loss": 0.2303,
       "step": 9472
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.423230355320531e-06,
-      "loss": 0.2527,
+      "loss": 0.0794,
       "step": 9473
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.4213748956303927e-06,
-      "loss": 0.2487,
+      "loss": 0.0804,
       "step": 9474
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.419519435940254e-06,
-      "loss": 0.1378,
+      "loss": 0.0647,
       "step": 9475
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.417663976250116e-06,
-      "loss": 0.1963,
+      "loss": 0.0403,
       "step": 9476
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.415808516559978e-06,
-      "loss": 0.0295,
+      "loss": 0.1083,
       "step": 9477
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.41395305686984e-06,
-      "loss": 0.0302,
+      "loss": 0.0199,
       "step": 9478
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.4120975971797012e-06,
-      "loss": 0.0297,
+      "loss": 0.0926,
       "step": 9479
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.410242137489563e-06,
-      "loss": 0.2517,
+      "loss": 0.0479,
       "step": 9480
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.408386677799425e-06,
-      "loss": 0.0831,
+      "loss": 0.041,
       "step": 9481
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.406531218109287e-06,
-      "loss": 0.0833,
+      "loss": 0.0152,
       "step": 9482
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.4046757584191487e-06,
-      "loss": 0.0875,
+      "loss": 0.1814,
       "step": 9483
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.40282029872901e-06,
-      "loss": 0.1389,
+      "loss": 0.0262,
       "step": 9484
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.400964839038872e-06,
-      "loss": 0.1907,
+      "loss": 0.1523,
       "step": 9485
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.399109379348734e-06,
-      "loss": 0.0293,
+      "loss": 0.0225,
       "step": 9486
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.3972539196585958e-06,
-      "loss": 0.1399,
+      "loss": 0.0191,
       "step": 9487
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.3953984599684572e-06,
-      "loss": 0.2495,
+      "loss": 0.1026,
       "step": 9488
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.393543000278319e-06,
-      "loss": 0.1939,
+      "loss": 0.0584,
       "step": 9489
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.391687540588181e-06,
-      "loss": 0.1389,
+      "loss": 0.1038,
       "step": 9490
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.389832080898043e-06,
-      "loss": 0.0292,
+      "loss": 0.1426,
       "step": 9491
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.3879766212079043e-06,
-      "loss": 0.1407,
+      "loss": 0.0158,
       "step": 9492
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.386121161517766e-06,
-      "loss": 0.0284,
+      "loss": 0.0695,
       "step": 9493
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.384265701827628e-06,
-      "loss": 0.2509,
+      "loss": 0.1201,
       "step": 9494
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.38241024213749e-06,
-      "loss": 0.1396,
+      "loss": 0.0634,
       "step": 9495
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.3805547824473513e-06,
-      "loss": 0.141,
+      "loss": 0.0138,
       "step": 9496
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.3786993227572132e-06,
-      "loss": 0.1433,
+      "loss": 0.0511,
       "step": 9497
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.376843863067075e-06,
-      "loss": 0.193,
+      "loss": 0.1255,
       "step": 9498
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.374988403376937e-06,
-      "loss": 0.0809,
+      "loss": 0.0987,
       "step": 9499
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.373132943686799e-06,
-      "loss": 0.3543,
+      "loss": 0.1437,
       "step": 9500
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.3712774839966603e-06,
-      "loss": 0.087,
+      "loss": 0.1188,
       "step": 9501
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.369422024306522e-06,
-      "loss": 0.0284,
+      "loss": 0.1088,
       "step": 9502
     },
     {
       "epoch": 2.64,
       "learning_rate": 2.367566564616384e-06,
-      "loss": 0.2437,
+      "loss": 0.1047,
       "step": 9503
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.365711104926246e-06,
-      "loss": 0.0291,
+      "loss": 0.0649,
       "step": 9504
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3638556452361073e-06,
-      "loss": 0.0292,
+      "loss": 0.0794,
       "step": 9505
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.362000185545969e-06,
-      "loss": 0.0291,
+      "loss": 0.1141,
       "step": 9506
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.360144725855831e-06,
-      "loss": 0.0289,
+      "loss": 0.0862,
       "step": 9507
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.358289266165693e-06,
-      "loss": 0.252,
+      "loss": 0.0863,
       "step": 9508
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3564338064755544e-06,
-      "loss": 0.142,
+      "loss": 0.0231,
       "step": 9509
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3545783467854163e-06,
-      "loss": 0.0828,
+      "loss": 0.1892,
       "step": 9510
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.352722887095278e-06,
-      "loss": 0.1389,
+      "loss": 0.0238,
       "step": 9511
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.35086742740514e-06,
-      "loss": 0.1953,
+      "loss": 0.1213,
       "step": 9512
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.349011967715002e-06,
-      "loss": 0.1944,
+      "loss": 0.0166,
       "step": 9513
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3471565080248633e-06,
-      "loss": 0.1957,
+      "loss": 0.0915,
       "step": 9514
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.345301048334725e-06,
-      "loss": 0.1934,
+      "loss": 0.0383,
       "step": 9515
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.343445588644587e-06,
-      "loss": 0.194,
+      "loss": 0.0607,
       "step": 9516
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.341590128954449e-06,
-      "loss": 0.1421,
+      "loss": 0.1361,
       "step": 9517
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3397346692643104e-06,
-      "loss": 0.0854,
+      "loss": 0.0337,
       "step": 9518
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3378792095741723e-06,
-      "loss": 0.1364,
+      "loss": 0.0563,
       "step": 9519
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.336023749884034e-06,
-      "loss": 0.0861,
+      "loss": 0.0653,
       "step": 9520
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.334168290193896e-06,
-      "loss": 0.2535,
+      "loss": 0.026,
       "step": 9521
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3323128305037574e-06,
-      "loss": 0.0816,
+      "loss": 0.0488,
       "step": 9522
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3304573708136193e-06,
-      "loss": 0.0856,
+      "loss": 0.0676,
       "step": 9523
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.328601911123481e-06,
-      "loss": 0.3068,
+      "loss": 0.0919,
       "step": 9524
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.326746451433343e-06,
-      "loss": 0.1929,
+      "loss": 0.1011,
       "step": 9525
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3248909917432045e-06,
-      "loss": 0.1929,
+      "loss": 0.015,
       "step": 9526
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3230355320530664e-06,
-      "loss": 0.2478,
+      "loss": 0.0267,
       "step": 9527
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.321180072362928e-06,
-      "loss": 0.3046,
+      "loss": 0.0538,
       "step": 9528
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3193246126727897e-06,
-      "loss": 0.0857,
+      "loss": 0.093,
       "step": 9529
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3174691529826516e-06,
-      "loss": 0.1417,
+      "loss": 0.0267,
       "step": 9530
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3156136932925134e-06,
-      "loss": 0.137,
+      "loss": 0.0191,
       "step": 9531
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.313758233602375e-06,
-      "loss": 0.0314,
+      "loss": 0.1319,
       "step": 9532
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3119027739122367e-06,
-      "loss": 0.1935,
+      "loss": 0.0911,
       "step": 9533
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3100473142220986e-06,
-      "loss": 0.0307,
+      "loss": 0.1581,
       "step": 9534
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3081918545319605e-06,
-      "loss": 0.1412,
+      "loss": 0.0519,
       "step": 9535
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3063363948418224e-06,
-      "loss": 0.1421,
+      "loss": 0.0453,
       "step": 9536
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.304480935151684e-06,
-      "loss": 0.0869,
+      "loss": 0.0713,
       "step": 9537
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3026254754615457e-06,
-      "loss": 0.1418,
+      "loss": 0.0551,
       "step": 9538
     },
     {
       "epoch": 2.65,
       "learning_rate": 2.3007700157714075e-06,
-      "loss": 0.0297,
+      "loss": 0.1203,
       "step": 9539
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2989145560812694e-06,
-      "loss": 0.1401,
+      "loss": 0.0701,
       "step": 9540
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.297059096391131e-06,
-      "loss": 0.0315,
+      "loss": 0.0225,
       "step": 9541
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2952036367009927e-06,
-      "loss": 0.1892,
+      "loss": 0.0243,
       "step": 9542
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2933481770108546e-06,
-      "loss": 0.2492,
+      "loss": 0.0683,
       "step": 9543
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2914927173207165e-06,
-      "loss": 0.0838,
+      "loss": 0.1414,
       "step": 9544
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.289637257630578e-06,
-      "loss": 0.1405,
+      "loss": 0.0756,
       "step": 9545
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.28778179794044e-06,
-      "loss": 0.0833,
+      "loss": 0.266,
       "step": 9546
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2859263382503017e-06,
-      "loss": 0.086,
+      "loss": 0.0446,
       "step": 9547
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2840708785601635e-06,
-      "loss": 0.0304,
+      "loss": 0.0544,
       "step": 9548
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.282215418870025e-06,
-      "loss": 0.1383,
+      "loss": 0.0402,
       "step": 9549
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.280359959179887e-06,
-      "loss": 0.192,
+      "loss": 0.1839,
       "step": 9550
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2785044994897487e-06,
-      "loss": 0.2982,
+      "loss": 0.0439,
       "step": 9551
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2766490397996106e-06,
-      "loss": 0.0876,
+      "loss": 0.126,
       "step": 9552
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2747935801094725e-06,
-      "loss": 0.1859,
+      "loss": 0.0141,
       "step": 9553
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.272938120419334e-06,
-      "loss": 0.252,
+      "loss": 0.0163,
       "step": 9554
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2710826607291958e-06,
-      "loss": 0.1977,
+      "loss": 0.0454,
       "step": 9555
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2692272010390577e-06,
-      "loss": 0.0839,
+      "loss": 0.0565,
       "step": 9556
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2673717413489195e-06,
-      "loss": 0.1906,
+      "loss": 0.102,
       "step": 9557
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.265516281658781e-06,
-      "loss": 0.0309,
+      "loss": 0.1176,
       "step": 9558
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.263660821968643e-06,
-      "loss": 0.0303,
+      "loss": 0.0276,
       "step": 9559
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2618053622785047e-06,
-      "loss": 0.0864,
+      "loss": 0.1508,
       "step": 9560
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2599499025883666e-06,
-      "loss": 0.2988,
+      "loss": 0.0521,
       "step": 9561
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.258094442898228e-06,
-      "loss": 0.1408,
+      "loss": 0.1282,
       "step": 9562
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.25623898320809e-06,
-      "loss": 0.1392,
+      "loss": 0.0597,
       "step": 9563
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2543835235179518e-06,
-      "loss": 0.1421,
+      "loss": 0.0227,
       "step": 9564
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2525280638278136e-06,
-      "loss": 0.1392,
+      "loss": 0.0399,
       "step": 9565
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2506726041376755e-06,
-      "loss": 0.1368,
+      "loss": 0.0667,
       "step": 9566
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.248817144447537e-06,
-      "loss": 0.0847,
+      "loss": 0.0159,
       "step": 9567
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.246961684757399e-06,
-      "loss": 0.1398,
+      "loss": 0.0938,
       "step": 9568
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2451062250672607e-06,
-      "loss": 0.0844,
+      "loss": 0.0543,
       "step": 9569
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2432507653771226e-06,
-      "loss": 0.0311,
+      "loss": 0.0122,
       "step": 9570
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.241395305686984e-06,
-      "loss": 0.0314,
+      "loss": 0.02,
       "step": 9571
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.239539845996846e-06,
-      "loss": 0.1402,
+      "loss": 0.0203,
       "step": 9572
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2376843863067078e-06,
-      "loss": 0.0868,
+      "loss": 0.1,
       "step": 9573
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.2358289266165696e-06,
-      "loss": 0.194,
+      "loss": 0.0085,
       "step": 9574
     },
     {
       "epoch": 2.66,
       "learning_rate": 2.233973466926431e-06,
-      "loss": 0.1397,
+      "loss": 0.0427,
       "step": 9575
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.232118007236293e-06,
-      "loss": 0.2458,
+      "loss": 0.0163,
       "step": 9576
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.230262547546155e-06,
-      "loss": 0.1923,
+      "loss": 0.0526,
       "step": 9577
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.2284070878560167e-06,
-      "loss": 0.1932,
+      "loss": 0.1316,
       "step": 9578
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.226551628165878e-06,
-      "loss": 0.1396,
+      "loss": 0.0497,
       "step": 9579
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.22469616847574e-06,
-      "loss": 0.0857,
+      "loss": 0.0492,
       "step": 9580
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.222840708785602e-06,
-      "loss": 0.1362,
+      "loss": 0.0434,
       "step": 9581
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.2209852490954637e-06,
-      "loss": 0.138,
+      "loss": 0.0448,
       "step": 9582
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.2191297894053256e-06,
-      "loss": 0.1927,
+      "loss": 0.0829,
       "step": 9583
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.217274329715187e-06,
-      "loss": 0.0822,
+      "loss": 0.1416,
       "step": 9584
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.215418870025049e-06,
-      "loss": 0.1881,
+      "loss": 0.0198,
       "step": 9585
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.213563410334911e-06,
-      "loss": 0.1437,
+      "loss": 0.0233,
       "step": 9586
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.2117079506447727e-06,
-      "loss": 0.14,
+      "loss": 0.0572,
       "step": 9587
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.209852490954634e-06,
-      "loss": 0.0303,
+      "loss": 0.0484,
       "step": 9588
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.207997031264496e-06,
-      "loss": 0.194,
+      "loss": 0.0541,
       "step": 9589
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.206141571574358e-06,
-      "loss": 0.1337,
+      "loss": 0.0556,
       "step": 9590
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.2042861118842197e-06,
-      "loss": 0.0858,
+      "loss": 0.0189,
       "step": 9591
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.202430652194081e-06,
-      "loss": 0.0848,
+      "loss": 0.1574,
       "step": 9592
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.200575192503943e-06,
-      "loss": 0.0864,
+      "loss": 0.0628,
       "step": 9593
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.198719732813805e-06,
-      "loss": 0.0851,
+      "loss": 0.0465,
       "step": 9594
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.196864273123667e-06,
-      "loss": 0.1383,
+      "loss": 0.0219,
       "step": 9595
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.1950088134335287e-06,
-      "loss": 0.031,
+      "loss": 0.0197,
       "step": 9596
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.19315335374339e-06,
-      "loss": 0.1392,
+      "loss": 0.0521,
       "step": 9597
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.191297894053252e-06,
-      "loss": 0.1362,
+      "loss": 0.0558,
       "step": 9598
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.189442434363114e-06,
-      "loss": 0.1871,
+      "loss": 0.2971,
       "step": 9599
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.1875869746729757e-06,
-      "loss": 0.3619,
+      "loss": 0.0631,
       "step": 9600
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.185731514982837e-06,
-      "loss": 0.1375,
+      "loss": 0.0672,
       "step": 9601
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.183876055292699e-06,
-      "loss": 0.0298,
+      "loss": 0.0399,
       "step": 9602
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.1820205956025605e-06,
-      "loss": 0.2453,
+      "loss": 0.0444,
       "step": 9603
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.1801651359124224e-06,
-      "loss": 0.2455,
+      "loss": 0.0172,
       "step": 9604
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.1783096762222842e-06,
-      "loss": 0.0316,
+      "loss": 0.1561,
       "step": 9605
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.176454216532146e-06,
-      "loss": 0.0305,
+      "loss": 0.016,
       "step": 9606
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.1745987568420075e-06,
-      "loss": 0.14,
+      "loss": 0.0909,
       "step": 9607
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.1727432971518694e-06,
-      "loss": 0.085,
+      "loss": 0.0152,
       "step": 9608
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.1708878374617313e-06,
-      "loss": 0.0849,
+      "loss": 0.1557,
       "step": 9609
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.169032377771593e-06,
-      "loss": 0.1939,
+      "loss": 0.0832,
       "step": 9610
     },
     {
       "epoch": 2.67,
       "learning_rate": 2.1671769180814546e-06,
-      "loss": 0.3073,
+      "loss": 0.0192,
       "step": 9611
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1653214583913165e-06,
-      "loss": 0.1388,
+      "loss": 0.0249,
       "step": 9612
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1634659987011783e-06,
-      "loss": 0.2495,
+      "loss": 0.0956,
       "step": 9613
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1616105390110402e-06,
-      "loss": 0.0826,
+      "loss": 0.0545,
       "step": 9614
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1597550793209017e-06,
-      "loss": 0.1901,
+      "loss": 0.013,
       "step": 9615
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1578996196307635e-06,
-      "loss": 0.084,
+      "loss": 0.1359,
       "step": 9616
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1560441599406254e-06,
-      "loss": 0.0848,
+      "loss": 0.0187,
       "step": 9617
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1541887002504873e-06,
-      "loss": 0.1949,
+      "loss": 0.0395,
       "step": 9618
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1523332405603487e-06,
-      "loss": 0.0315,
+      "loss": 0.097,
       "step": 9619
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1504777808702106e-06,
-      "loss": 0.1355,
+      "loss": 0.0562,
       "step": 9620
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1486223211800725e-06,
-      "loss": 0.1376,
+      "loss": 0.0154,
       "step": 9621
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1467668614899343e-06,
-      "loss": 0.0855,
+      "loss": 0.0746,
       "step": 9622
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.144911401799796e-06,
-      "loss": 0.085,
+      "loss": 0.0493,
       "step": 9623
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1430559421096577e-06,
-      "loss": 0.1401,
+      "loss": 0.1516,
       "step": 9624
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1412004824195195e-06,
-      "loss": 0.0306,
+      "loss": 0.0654,
       "step": 9625
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1393450227293814e-06,
-      "loss": 0.1385,
+      "loss": 0.0134,
       "step": 9626
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1374895630392433e-06,
-      "loss": 0.0855,
+      "loss": 0.1004,
       "step": 9627
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1356341033491047e-06,
-      "loss": 0.2548,
+      "loss": 0.0104,
       "step": 9628
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1337786436589666e-06,
-      "loss": 0.1946,
+      "loss": 0.0954,
       "step": 9629
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1319231839688285e-06,
-      "loss": 0.0842,
+      "loss": 0.0726,
       "step": 9630
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1300677242786903e-06,
-      "loss": 0.0297,
+      "loss": 0.0385,
       "step": 9631
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1282122645885518e-06,
-      "loss": 0.3581,
+      "loss": 0.0135,
       "step": 9632
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1263568048984136e-06,
-      "loss": 0.1416,
+      "loss": 0.0578,
       "step": 9633
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1245013452082755e-06,
-      "loss": 0.1411,
+      "loss": 0.1547,
       "step": 9634
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1226458855181374e-06,
-      "loss": 0.0862,
+      "loss": 0.0133,
       "step": 9635
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1207904258279993e-06,
-      "loss": 0.1923,
+      "loss": 0.014,
       "step": 9636
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1189349661378607e-06,
-      "loss": 0.1414,
+      "loss": 0.0534,
       "step": 9637
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1170795064477226e-06,
-      "loss": 0.1411,
+      "loss": 0.1177,
       "step": 9638
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1152240467575844e-06,
-      "loss": 0.0305,
+      "loss": 0.0253,
       "step": 9639
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1133685870674463e-06,
-      "loss": 0.1402,
+      "loss": 0.0072,
       "step": 9640
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1115131273773078e-06,
-      "loss": 0.1956,
+      "loss": 0.018,
       "step": 9641
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1096576676871696e-06,
-      "loss": 0.0902,
+      "loss": 0.0501,
       "step": 9642
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1078022079970315e-06,
-      "loss": 0.1381,
+      "loss": 0.0775,
       "step": 9643
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1059467483068934e-06,
-      "loss": 0.1409,
+      "loss": 0.0641,
       "step": 9644
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.104091288616755e-06,
-      "loss": 0.1416,
+      "loss": 0.0814,
       "step": 9645
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1022358289266167e-06,
-      "loss": 0.1382,
+      "loss": 0.0128,
       "step": 9646
     },
     {
       "epoch": 2.68,
       "learning_rate": 2.1003803692364786e-06,
-      "loss": 0.1411,
+      "loss": 0.055,
       "step": 9647
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0985249095463404e-06,
-      "loss": 0.3087,
+      "loss": 0.0496,
       "step": 9648
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.096669449856202e-06,
-      "loss": 0.1963,
+      "loss": 0.0759,
       "step": 9649
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0948139901660637e-06,
-      "loss": 0.2458,
+      "loss": 0.0228,
       "step": 9650
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0929585304759256e-06,
-      "loss": 0.1395,
+      "loss": 0.1987,
       "step": 9651
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0911030707857875e-06,
-      "loss": 0.0853,
+      "loss": 0.0912,
       "step": 9652
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0892476110956494e-06,
-      "loss": 0.0308,
+      "loss": 0.0124,
       "step": 9653
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.087392151405511e-06,
-      "loss": 0.0843,
+      "loss": 0.0644,
       "step": 9654
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0855366917153727e-06,
-      "loss": 0.0312,
+      "loss": 0.0807,
       "step": 9655
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0836812320252345e-06,
-      "loss": 0.3019,
+      "loss": 0.0247,
       "step": 9656
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0818257723350964e-06,
-      "loss": 0.1954,
+      "loss": 0.118,
       "step": 9657
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.079970312644958e-06,
-      "loss": 0.3046,
+      "loss": 0.0657,
       "step": 9658
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0781148529548197e-06,
-      "loss": 0.0861,
+      "loss": 0.0115,
       "step": 9659
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0762593932646816e-06,
-      "loss": 0.1408,
+      "loss": 0.0595,
       "step": 9660
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0744039335745435e-06,
-      "loss": 0.1391,
+      "loss": 0.1379,
       "step": 9661
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.072548473884405e-06,
-      "loss": 0.2544,
+      "loss": 0.0445,
       "step": 9662
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.070693014194267e-06,
-      "loss": 0.2998,
+      "loss": 0.0513,
       "step": 9663
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0688375545041287e-06,
-      "loss": 0.0836,
+      "loss": 0.0547,
       "step": 9664
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0669820948139905e-06,
-      "loss": 0.1951,
+      "loss": 0.0114,
       "step": 9665
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0651266351238524e-06,
-      "loss": 0.0321,
+      "loss": 0.1387,
       "step": 9666
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.063271175433714e-06,
-      "loss": 0.1896,
+      "loss": 0.0514,
       "step": 9667
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0614157157435757e-06,
-      "loss": 0.0866,
+      "loss": 0.0665,
       "step": 9668
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0595602560534376e-06,
-      "loss": 0.0318,
+      "loss": 0.0185,
       "step": 9669
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0577047963632995e-06,
-      "loss": 0.0852,
+      "loss": 0.044,
       "step": 9670
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.055849336673161e-06,
-      "loss": 0.143,
+      "loss": 0.0345,
       "step": 9671
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0539938769830228e-06,
-      "loss": 0.1407,
+      "loss": 0.0239,
       "step": 9672
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0521384172928846e-06,
-      "loss": 0.0861,
+      "loss": 0.0231,
       "step": 9673
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0502829576027465e-06,
-      "loss": 0.1929,
+      "loss": 0.1225,
       "step": 9674
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.048427497912608e-06,
-      "loss": 0.0329,
+      "loss": 0.0749,
       "step": 9675
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.04657203822247e-06,
-      "loss": 0.1417,
+      "loss": 0.0131,
       "step": 9676
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0447165785323317e-06,
-      "loss": 0.1944,
+      "loss": 0.0983,
       "step": 9677
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0428611188421936e-06,
-      "loss": 0.1954,
+      "loss": 0.0442,
       "step": 9678
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.041005659152055e-06,
-      "loss": 0.3064,
+      "loss": 0.0185,
       "step": 9679
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.039150199461917e-06,
-      "loss": 0.1415,
+      "loss": 0.0972,
       "step": 9680
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0372947397717783e-06,
-      "loss": 0.1953,
+      "loss": 0.2024,
       "step": 9681
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.0354392800816402e-06,
-      "loss": 0.1399,
+      "loss": 0.0166,
       "step": 9682
     },
     {
       "epoch": 2.69,
       "learning_rate": 2.033583820391502e-06,
-      "loss": 0.1378,
+      "loss": 0.0573,
       "step": 9683
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.031728360701364e-06,
-      "loss": 0.1872,
+      "loss": 0.0866,
       "step": 9684
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.0298729010112254e-06,
-      "loss": 0.1941,
+      "loss": 0.0172,
       "step": 9685
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.0280174413210873e-06,
-      "loss": 0.0889,
+      "loss": 0.0163,
       "step": 9686
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.026161981630949e-06,
-      "loss": 0.0323,
+      "loss": 0.1116,
       "step": 9687
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.024306521940811e-06,
-      "loss": 0.0851,
+      "loss": 0.0093,
       "step": 9688
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.0224510622506725e-06,
-      "loss": 0.2447,
+      "loss": 0.0718,
       "step": 9689
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.0205956025605343e-06,
-      "loss": 0.1907,
+      "loss": 0.107,
       "step": 9690
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.018740142870396e-06,
-      "loss": 0.0848,
+      "loss": 0.0201,
       "step": 9691
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.016884683180258e-06,
-      "loss": 0.0882,
+      "loss": 0.0535,
       "step": 9692
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.01502922349012e-06,
-      "loss": 0.087,
+      "loss": 0.0116,
       "step": 9693
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.0131737637999814e-06,
-      "loss": 0.1943,
+      "loss": 0.0112,
       "step": 9694
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.0113183041098433e-06,
-      "loss": 0.1384,
+      "loss": 0.0705,
       "step": 9695
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.009462844419705e-06,
-      "loss": 0.1378,
+      "loss": 0.1876,
       "step": 9696
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.007607384729567e-06,
-      "loss": 0.1922,
+      "loss": 0.0117,
       "step": 9697
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.0057519250394284e-06,
-      "loss": 0.0827,
+      "loss": 0.0839,
       "step": 9698
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.0038964653492903e-06,
-      "loss": 0.2434,
+      "loss": 0.0824,
       "step": 9699
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.002041005659152e-06,
-      "loss": 0.1441,
+      "loss": 0.0674,
       "step": 9700
     },
     {
       "epoch": 2.7,
       "learning_rate": 2.000185545969014e-06,
-      "loss": 0.138,
+      "loss": 0.0145,
       "step": 9701
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.9983300862788755e-06,
-      "loss": 0.0881,
+      "loss": 0.0061,
       "step": 9702
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.9964746265887374e-06,
-      "loss": 0.1397,
+      "loss": 0.0856,
       "step": 9703
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.9946191668985992e-06,
-      "loss": 0.1419,
+      "loss": 0.0096,
       "step": 9704
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.992763707208461e-06,
-      "loss": 0.244,
+      "loss": 0.0131,
       "step": 9705
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.990908247518323e-06,
-      "loss": 0.0327,
+      "loss": 0.0455,
       "step": 9706
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.9890527878281844e-06,
-      "loss": 0.1339,
+      "loss": 0.1068,
       "step": 9707
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.9871973281380463e-06,
-      "loss": 0.1394,
+      "loss": 0.1438,
       "step": 9708
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.985341868447908e-06,
-      "loss": 0.1384,
+      "loss": 0.0057,
       "step": 9709
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.98348640875777e-06,
-      "loss": 0.2499,
+      "loss": 0.0593,
       "step": 9710
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.9816309490676315e-06,
-      "loss": 0.1952,
+      "loss": 0.0886,
       "step": 9711
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.9797754893774934e-06,
-      "loss": 0.1407,
+      "loss": 0.1304,
       "step": 9712
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.9779200296873552e-06,
-      "loss": 0.0848,
+      "loss": 0.0657,
       "step": 9713
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.976064569997217e-06,
-      "loss": 0.247,
+      "loss": 0.0714,
       "step": 9714
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.9742091103070786e-06,
-      "loss": 0.0855,
+      "loss": 0.0755,
       "step": 9715
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.9723536506169404e-06,
-      "loss": 0.1934,
+      "loss": 0.015,
       "step": 9716
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.9704981909268023e-06,
-      "loss": 0.2445,
+      "loss": 0.0554,
       "step": 9717
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.968642731236664e-06,
-      "loss": 0.14,
+      "loss": 0.0426,
       "step": 9718
     },
     {
       "epoch": 2.7,
       "learning_rate": 1.9667872715465256e-06,
-      "loss": 0.1923,
+      "loss": 0.018,
       "step": 9719
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9649318118563875e-06,
-      "loss": 0.0874,
+      "loss": 0.0181,
       "step": 9720
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9630763521662494e-06,
-      "loss": 0.2983,
+      "loss": 0.092,
       "step": 9721
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9612208924761112e-06,
-      "loss": 0.0876,
+      "loss": 0.019,
       "step": 9722
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.959365432785973e-06,
-      "loss": 0.1419,
+      "loss": 0.0494,
       "step": 9723
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9575099730958345e-06,
-      "loss": 0.2464,
+      "loss": 0.0148,
       "step": 9724
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9556545134056964e-06,
-      "loss": 0.2449,
+      "loss": 0.0829,
       "step": 9725
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9537990537155583e-06,
-      "loss": 0.1383,
+      "loss": 0.1581,
       "step": 9726
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.95194359402542e-06,
-      "loss": 0.0859,
+      "loss": 0.0426,
       "step": 9727
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9500881343352816e-06,
-      "loss": 0.0874,
+      "loss": 0.0541,
       "step": 9728
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9482326746451435e-06,
-      "loss": 0.0854,
+      "loss": 0.2004,
       "step": 9729
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9463772149550053e-06,
-      "loss": 0.3559,
+      "loss": 0.046,
       "step": 9730
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9445217552648672e-06,
-      "loss": 0.1422,
+      "loss": 0.0189,
       "step": 9731
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9426662955747287e-06,
-      "loss": 0.0845,
+      "loss": 0.1353,
       "step": 9732
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9408108358845905e-06,
-      "loss": 0.246,
+      "loss": 0.1362,
       "step": 9733
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9389553761944524e-06,
-      "loss": 0.2995,
+      "loss": 0.0532,
       "step": 9734
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9370999165043143e-06,
-      "loss": 0.0872,
+      "loss": 0.0428,
       "step": 9735
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.935244456814176e-06,
-      "loss": 0.191,
+      "loss": 0.0681,
       "step": 9736
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9333889971240376e-06,
-      "loss": 0.1432,
+      "loss": 0.0484,
       "step": 9737
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9315335374338995e-06,
-      "loss": 0.0867,
+      "loss": 0.013,
       "step": 9738
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9296780777437613e-06,
-      "loss": 0.1374,
+      "loss": 0.0099,
       "step": 9739
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.927822618053623e-06,
-      "loss": 0.2438,
+      "loss": 0.0468,
       "step": 9740
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9259671583634846e-06,
-      "loss": 0.0867,
+      "loss": 0.0433,
       "step": 9741
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9241116986733465e-06,
-      "loss": 0.1398,
+      "loss": 0.0182,
       "step": 9742
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9222562389832084e-06,
-      "loss": 0.292,
+      "loss": 0.017,
       "step": 9743
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9204007792930703e-06,
-      "loss": 0.1397,
+      "loss": 0.0559,
       "step": 9744
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9185453196029317e-06,
-      "loss": 0.0346,
+      "loss": 0.0982,
       "step": 9745
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9166898599127936e-06,
-      "loss": 0.193,
+      "loss": 0.013,
       "step": 9746
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9148344002226554e-06,
-      "loss": 0.2464,
+      "loss": 0.0409,
       "step": 9747
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9129789405325173e-06,
-      "loss": 0.0848,
+      "loss": 0.0607,
       "step": 9748
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9111234808423788e-06,
-      "loss": 0.1393,
+      "loss": 0.1555,
       "step": 9749
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9092680211522406e-06,
-      "loss": 0.1422,
+      "loss": 0.2465,
       "step": 9750
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9074125614621025e-06,
-      "loss": 0.1941,
+      "loss": 0.0611,
       "step": 9751
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.9055571017719644e-06,
-      "loss": 0.2408,
+      "loss": 0.0208,
       "step": 9752
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.903701642081826e-06,
-      "loss": 0.0851,
+      "loss": 0.0122,
       "step": 9753
     },
     {
       "epoch": 2.71,
       "learning_rate": 1.901846182391688e-06,
-      "loss": 0.188,
+      "loss": 0.0522,
       "step": 9754
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8999907227015494e-06,
-      "loss": 0.0873,
+      "loss": 0.0479,
       "step": 9755
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.898135263011411e-06,
-      "loss": 0.0886,
+      "loss": 0.0562,
       "step": 9756
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8962798033212729e-06,
-      "loss": 0.0906,
+      "loss": 0.0456,
       "step": 9757
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8944243436311345e-06,
-      "loss": 0.1385,
+      "loss": 0.1891,
       "step": 9758
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8925688839409964e-06,
-      "loss": 0.0343,
+      "loss": 0.0145,
       "step": 9759
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8907134242508583e-06,
-      "loss": 0.0341,
+      "loss": 0.0084,
       "step": 9760
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.88885796456072e-06,
-      "loss": 0.0869,
+      "loss": 0.0101,
       "step": 9761
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8870025048705818e-06,
-      "loss": 0.2517,
+      "loss": 0.0239,
       "step": 9762
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8851470451804435e-06,
-      "loss": 0.0846,
+      "loss": 0.1026,
       "step": 9763
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8832915854903053e-06,
-      "loss": 0.0349,
+      "loss": 0.0151,
       "step": 9764
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.881436125800167e-06,
-      "loss": 0.2458,
+      "loss": 0.01,
       "step": 9765
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8795806661100289e-06,
-      "loss": 0.0355,
+      "loss": 0.0872,
       "step": 9766
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8777252064198905e-06,
-      "loss": 0.1413,
+      "loss": 0.0602,
       "step": 9767
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8758697467297524e-06,
-      "loss": 0.1888,
+      "loss": 0.0735,
       "step": 9768
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.874014287039614e-06,
-      "loss": 0.2928,
+      "loss": 0.1058,
       "step": 9769
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.872158827349476e-06,
-      "loss": 0.0865,
+      "loss": 0.1295,
       "step": 9770
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8703033676593376e-06,
-      "loss": 0.0857,
+      "loss": 0.0471,
       "step": 9771
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8684479079691995e-06,
-      "loss": 0.1375,
+      "loss": 0.0802,
       "step": 9772
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8665924482790611e-06,
-      "loss": 0.035,
+      "loss": 0.1044,
       "step": 9773
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.864736988588923e-06,
-      "loss": 0.3493,
+      "loss": 0.0603,
       "step": 9774
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8628815288987849e-06,
-      "loss": 0.1951,
+      "loss": 0.0342,
       "step": 9775
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8610260692086465e-06,
-      "loss": 0.0345,
+      "loss": 0.013,
       "step": 9776
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8591706095185084e-06,
-      "loss": 0.0883,
+      "loss": 0.1205,
       "step": 9777
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.85731514982837e-06,
-      "loss": 0.0346,
+      "loss": 0.0211,
       "step": 9778
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.855459690138232e-06,
-      "loss": 0.2437,
+      "loss": 0.0824,
       "step": 9779
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8536042304480936e-06,
-      "loss": 0.0874,
+      "loss": 0.0184,
       "step": 9780
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8517487707579554e-06,
-      "loss": 0.0343,
+      "loss": 0.0598,
       "step": 9781
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8498933110678171e-06,
-      "loss": 0.1365,
+      "loss": 0.0166,
       "step": 9782
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.848037851377679e-06,
-      "loss": 0.0884,
+      "loss": 0.0108,
       "step": 9783
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8461823916875406e-06,
-      "loss": 0.1428,
+      "loss": 0.1402,
       "step": 9784
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8443269319974025e-06,
-      "loss": 0.1947,
+      "loss": 0.064,
       "step": 9785
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8424714723072642e-06,
-      "loss": 0.1389,
+      "loss": 0.0103,
       "step": 9786
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.840616012617126e-06,
-      "loss": 0.1923,
+      "loss": 0.1969,
       "step": 9787
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8387605529269877e-06,
-      "loss": 0.137,
+      "loss": 0.0167,
       "step": 9788
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8369050932368496e-06,
-      "loss": 0.0885,
+      "loss": 0.0158,
       "step": 9789
     },
     {
       "epoch": 2.72,
       "learning_rate": 1.8350496335467114e-06,
-      "loss": 0.0864,
+      "loss": 0.1802,
       "step": 9790
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.833194173856573e-06,
-      "loss": 0.2456,
+      "loss": 0.113,
       "step": 9791
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.831338714166435e-06,
-      "loss": 0.1391,
+      "loss": 0.0652,
       "step": 9792
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8294832544762966e-06,
-      "loss": 0.089,
+      "loss": 0.1921,
       "step": 9793
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8276277947861585e-06,
-      "loss": 0.0875,
+      "loss": 0.0215,
       "step": 9794
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8257723350960202e-06,
-      "loss": 0.0847,
+      "loss": 0.0484,
       "step": 9795
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.823916875405882e-06,
-      "loss": 0.0839,
+      "loss": 0.042,
       "step": 9796
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8220614157157437e-06,
-      "loss": 0.249,
+      "loss": 0.058,
       "step": 9797
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8202059560256056e-06,
-      "loss": 0.1428,
+      "loss": 0.0475,
       "step": 9798
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8183504963354672e-06,
-      "loss": 0.1891,
+      "loss": 0.0787,
       "step": 9799
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.816495036645329e-06,
-      "loss": 0.0333,
+      "loss": 0.0822,
       "step": 9800
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8146395769551907e-06,
-      "loss": 0.1372,
+      "loss": 0.1184,
       "step": 9801
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8127841172650526e-06,
-      "loss": 0.0864,
+      "loss": 0.0506,
       "step": 9802
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8109286575749143e-06,
-      "loss": 0.1365,
+      "loss": 0.0176,
       "step": 9803
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8090731978847761e-06,
-      "loss": 0.1419,
+      "loss": 0.0906,
       "step": 9804
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.807217738194638e-06,
-      "loss": 0.0871,
+      "loss": 0.0111,
       "step": 9805
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8053622785044997e-06,
-      "loss": 0.189,
+      "loss": 0.0469,
       "step": 9806
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8035068188143615e-06,
-      "loss": 0.1391,
+      "loss": 0.0198,
       "step": 9807
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.8016513591242232e-06,
-      "loss": 0.1366,
+      "loss": 0.1264,
       "step": 9808
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.799795899434085e-06,
-      "loss": 0.2445,
+      "loss": 0.0355,
       "step": 9809
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7979404397439467e-06,
-      "loss": 0.2438,
+      "loss": 0.0534,
       "step": 9810
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7960849800538086e-06,
-      "loss": 0.0336,
+      "loss": 0.0408,
       "step": 9811
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7942295203636703e-06,
-      "loss": 0.1944,
+      "loss": 0.0187,
       "step": 9812
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7923740606735321e-06,
-      "loss": 0.1399,
+      "loss": 0.0063,
       "step": 9813
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7905186009833938e-06,
-      "loss": 0.0336,
+      "loss": 0.0495,
       "step": 9814
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7886631412932557e-06,
-      "loss": 0.1402,
+      "loss": 0.1234,
       "step": 9815
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7868076816031173e-06,
-      "loss": 0.0883,
+      "loss": 0.0793,
       "step": 9816
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7849522219129792e-06,
-      "loss": 0.1381,
+      "loss": 0.0467,
       "step": 9817
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7830967622228408e-06,
-      "loss": 0.1954,
+      "loss": 0.142,
       "step": 9818
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7812413025327027e-06,
-      "loss": 0.1375,
+      "loss": 0.0722,
       "step": 9819
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7793858428425646e-06,
-      "loss": 0.1986,
+      "loss": 0.0511,
       "step": 9820
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7775303831524262e-06,
-      "loss": 0.0325,
+      "loss": 0.0154,
       "step": 9821
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7756749234622881e-06,
-      "loss": 0.2397,
+      "loss": 0.0169,
       "step": 9822
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7738194637721498e-06,
-      "loss": 0.0839,
+      "loss": 0.0925,
       "step": 9823
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7719640040820116e-06,
-      "loss": 0.0316,
+      "loss": 0.0201,
       "step": 9824
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7701085443918733e-06,
-      "loss": 0.2478,
+      "loss": 0.0807,
       "step": 9825
     },
     {
       "epoch": 2.73,
       "learning_rate": 1.7682530847017352e-06,
-      "loss": 0.1919,
+      "loss": 0.0097,
       "step": 9826
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7663976250115968e-06,
-      "loss": 0.0336,
+      "loss": 0.0632,
       "step": 9827
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7645421653214587e-06,
-      "loss": 0.0851,
+      "loss": 0.0596,
       "step": 9828
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7626867056313204e-06,
-      "loss": 0.2433,
+      "loss": 0.0426,
       "step": 9829
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7608312459411822e-06,
-      "loss": 0.0322,
+      "loss": 0.0213,
       "step": 9830
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7589757862510437e-06,
-      "loss": 0.0841,
+      "loss": 0.1043,
       "step": 9831
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7571203265609056e-06,
-      "loss": 0.1403,
+      "loss": 0.1198,
       "step": 9832
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7552648668707672e-06,
-      "loss": 0.2429,
+      "loss": 0.0489,
       "step": 9833
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.753409407180629e-06,
-      "loss": 0.0846,
+      "loss": 0.1004,
       "step": 9834
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7515539474904907e-06,
-      "loss": 0.0856,
+      "loss": 0.1127,
       "step": 9835
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7496984878003526e-06,
-      "loss": 0.1936,
+      "loss": 0.2209,
       "step": 9836
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7478430281102143e-06,
-      "loss": 0.1417,
+      "loss": 0.1532,
       "step": 9837
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7459875684200761e-06,
-      "loss": 0.0317,
+      "loss": 0.0575,
       "step": 9838
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7441321087299378e-06,
-      "loss": 0.1376,
+      "loss": 0.1165,
       "step": 9839
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7422766490397997e-06,
-      "loss": 0.0844,
+      "loss": 0.012,
       "step": 9840
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7404211893496613e-06,
-      "loss": 0.0847,
+      "loss": 0.0656,
       "step": 9841
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7385657296595232e-06,
-      "loss": 0.0849,
+      "loss": 0.1481,
       "step": 9842
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.736710269969385e-06,
-      "loss": 0.1959,
+      "loss": 0.0587,
       "step": 9843
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7348548102792467e-06,
-      "loss": 0.1925,
+      "loss": 0.0938,
       "step": 9844
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7329993505891086e-06,
-      "loss": 0.3454,
+      "loss": 0.0486,
       "step": 9845
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7311438908989703e-06,
-      "loss": 0.0878,
+      "loss": 0.0586,
       "step": 9846
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7292884312088321e-06,
-      "loss": 0.191,
+      "loss": 0.0564,
       "step": 9847
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7274329715186938e-06,
-      "loss": 0.1378,
+      "loss": 0.1236,
       "step": 9848
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7255775118285557e-06,
-      "loss": 0.1926,
+      "loss": 0.0602,
       "step": 9849
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7237220521384173e-06,
-      "loss": 0.1387,
+      "loss": 0.0739,
       "step": 9850
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7218665924482792e-06,
-      "loss": 0.1416,
+      "loss": 0.1007,
       "step": 9851
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7200111327581408e-06,
-      "loss": 0.1407,
+      "loss": 0.0521,
       "step": 9852
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7181556730680027e-06,
-      "loss": 0.1917,
+      "loss": 0.098,
       "step": 9853
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7163002133778644e-06,
-      "loss": 0.0842,
+      "loss": 0.0532,
       "step": 9854
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7144447536877262e-06,
-      "loss": 0.0874,
+      "loss": 0.0981,
       "step": 9855
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.712589293997588e-06,
-      "loss": 0.2517,
+      "loss": 0.0586,
       "step": 9856
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7107338343074498e-06,
-      "loss": 0.1396,
+      "loss": 0.1212,
       "step": 9857
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7088783746173116e-06,
-      "loss": 0.1382,
+      "loss": 0.0476,
       "step": 9858
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7070229149271733e-06,
-      "loss": 0.1953,
+      "loss": 0.0264,
       "step": 9859
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7051674552370352e-06,
-      "loss": 0.0325,
+      "loss": 0.0239,
       "step": 9860
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7033119955468968e-06,
-      "loss": 0.1403,
+      "loss": 0.0131,
       "step": 9861
     },
     {
       "epoch": 2.74,
       "learning_rate": 1.7014565358567587e-06,
-      "loss": 0.2462,
+      "loss": 0.0438,
       "step": 9862
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6996010761666204e-06,
-      "loss": 0.0889,
+      "loss": 0.0622,
       "step": 9863
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6977456164764822e-06,
-      "loss": 0.1901,
+      "loss": 0.0158,
       "step": 9864
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.695890156786344e-06,
-      "loss": 0.1943,
+      "loss": 0.0977,
       "step": 9865
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6940346970962058e-06,
-      "loss": 0.0832,
+      "loss": 0.0526,
       "step": 9866
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6921792374060674e-06,
-      "loss": 0.1953,
+      "loss": 0.0128,
       "step": 9867
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6903237777159293e-06,
-      "loss": 0.2437,
+      "loss": 0.1272,
       "step": 9868
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.688468318025791e-06,
-      "loss": 0.0861,
+      "loss": 0.0481,
       "step": 9869
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6866128583356528e-06,
-      "loss": 0.3014,
+      "loss": 0.0137,
       "step": 9870
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6847573986455145e-06,
-      "loss": 0.243,
+      "loss": 0.092,
       "step": 9871
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6829019389553764e-06,
-      "loss": 0.0323,
+      "loss": 0.0551,
       "step": 9872
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6810464792652382e-06,
-      "loss": 0.0327,
+      "loss": 0.0134,
       "step": 9873
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6791910195750999e-06,
-      "loss": 0.1912,
+      "loss": 0.0538,
       "step": 9874
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6773355598849618e-06,
-      "loss": 0.0882,
+      "loss": 0.1241,
       "step": 9875
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6754801001948234e-06,
-      "loss": 0.0843,
+      "loss": 0.0472,
       "step": 9876
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6736246405046853e-06,
-      "loss": 0.1387,
+      "loss": 0.0179,
       "step": 9877
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.671769180814547e-06,
-      "loss": 0.1907,
+      "loss": 0.0213,
       "step": 9878
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6699137211244088e-06,
-      "loss": 0.1414,
+      "loss": 0.0546,
       "step": 9879
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6680582614342705e-06,
-      "loss": 0.0877,
+      "loss": 0.0362,
       "step": 9880
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6662028017441323e-06,
-      "loss": 0.1405,
+      "loss": 0.0531,
       "step": 9881
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.664347342053994e-06,
-      "loss": 0.0835,
+      "loss": 0.0205,
       "step": 9882
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6624918823638559e-06,
-      "loss": 0.1412,
+      "loss": 0.0451,
       "step": 9883
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6606364226737175e-06,
-      "loss": 0.0864,
+      "loss": 0.0159,
       "step": 9884
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6587809629835794e-06,
-      "loss": 0.1922,
+      "loss": 0.0941,
       "step": 9885
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.656925503293441e-06,
-      "loss": 0.0326,
+      "loss": 0.0377,
       "step": 9886
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.655070043603303e-06,
-      "loss": 0.1387,
+      "loss": 0.0112,
       "step": 9887
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6532145839131648e-06,
-      "loss": 0.1412,
+      "loss": 0.0173,
       "step": 9888
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6513591242230265e-06,
-      "loss": 0.2497,
+      "loss": 0.0173,
       "step": 9889
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6495036645328883e-06,
-      "loss": 0.086,
+      "loss": 0.0989,
       "step": 9890
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.64764820484275e-06,
-      "loss": 0.0883,
+      "loss": 0.0128,
       "step": 9891
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6457927451526119e-06,
-      "loss": 0.0866,
+      "loss": 0.0127,
       "step": 9892
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6439372854624735e-06,
-      "loss": 0.1388,
+      "loss": 0.0455,
       "step": 9893
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6420818257723354e-06,
-      "loss": 0.0855,
+      "loss": 0.0506,
       "step": 9894
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.640226366082197e-06,
-      "loss": 0.1927,
+      "loss": 0.0455,
       "step": 9895
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.638370906392059e-06,
-      "loss": 0.0851,
+      "loss": 0.1903,
       "step": 9896
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6365154467019206e-06,
-      "loss": 0.1419,
+      "loss": 0.0199,
       "step": 9897
     },
     {
       "epoch": 2.75,
       "learning_rate": 1.6346599870117824e-06,
-      "loss": 0.1934,
+      "loss": 0.0814,
       "step": 9898
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.632804527321644e-06,
-      "loss": 0.1928,
+      "loss": 0.0098,
       "step": 9899
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.630949067631506e-06,
-      "loss": 0.1383,
+      "loss": 0.0136,
       "step": 9900
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.6290936079413676e-06,
-      "loss": 0.1953,
+      "loss": 0.0441,
       "step": 9901
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.6272381482512295e-06,
-      "loss": 0.0856,
+      "loss": 0.0561,
       "step": 9902
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.6253826885610914e-06,
-      "loss": 0.1441,
+      "loss": 0.093,
       "step": 9903
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.623527228870953e-06,
-      "loss": 0.1929,
+      "loss": 0.0674,
       "step": 9904
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.621671769180815e-06,
-      "loss": 0.2429,
+      "loss": 0.0113,
       "step": 9905
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.6198163094906766e-06,
-      "loss": 0.0854,
+      "loss": 0.0208,
       "step": 9906
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.617960849800538e-06,
-      "loss": 0.087,
+      "loss": 0.0447,
       "step": 9907
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.6161053901103999e-06,
-      "loss": 0.0837,
+      "loss": 0.0133,
       "step": 9908
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.6142499304202615e-06,
-      "loss": 0.1359,
+      "loss": 0.0495,
       "step": 9909
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.6123944707301234e-06,
-      "loss": 0.1923,
+      "loss": 0.1472,
       "step": 9910
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.610539011039985e-06,
-      "loss": 0.0315,
+      "loss": 0.1216,
       "step": 9911
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.608683551349847e-06,
-      "loss": 0.1407,
+      "loss": 0.0128,
       "step": 9912
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.6068280916597088e-06,
-      "loss": 0.032,
+      "loss": 0.0362,
       "step": 9913
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.6049726319695705e-06,
-      "loss": 0.033,
+      "loss": 0.0099,
       "step": 9914
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.6031171722794323e-06,
-      "loss": 0.0853,
+      "loss": 0.105,
       "step": 9915
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.601261712589294e-06,
-      "loss": 0.1402,
+      "loss": 0.0566,
       "step": 9916
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.5994062528991559e-06,
-      "loss": 0.135,
+      "loss": 0.1104,
       "step": 9917
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.5975507932090175e-06,
-      "loss": 0.139,
+      "loss": 0.0368,
       "step": 9918
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.5956953335188794e-06,
-      "loss": 0.0873,
+      "loss": 0.103,
       "step": 9919
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.593839873828741e-06,
-      "loss": 0.195,
+      "loss": 0.023,
       "step": 9920
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.591984414138603e-06,
-      "loss": 0.0883,
+      "loss": 0.0123,
       "step": 9921
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.5901289544484646e-06,
-      "loss": 0.2956,
+      "loss": 0.0575,
       "step": 9922
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.5882734947583265e-06,
-      "loss": 0.0316,
+      "loss": 0.1447,
       "step": 9923
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.5864180350681881e-06,
-      "loss": 0.086,
+      "loss": 0.0112,
       "step": 9924
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.58456257537805e-06,
-      "loss": 0.0834,
+      "loss": 0.0932,
       "step": 9925
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.5827071156879116e-06,
-      "loss": 0.1952,
+      "loss": 0.06,
       "step": 9926
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.5808516559977735e-06,
-      "loss": 0.1388,
+      "loss": 0.014,
       "step": 9927
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.5789961963076354e-06,
-      "loss": 0.1407,
+      "loss": 0.0144,
       "step": 9928
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.577140736617497e-06,
-      "loss": 0.0808,
+      "loss": 0.2019,
       "step": 9929
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.575285276927359e-06,
-      "loss": 0.1386,
+      "loss": 0.0698,
       "step": 9930
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.5734298172372206e-06,
-      "loss": 0.0868,
+      "loss": 0.0373,
       "step": 9931
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.5715743575470824e-06,
-      "loss": 0.1376,
+      "loss": 0.0489,
       "step": 9932
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.569718897856944e-06,
-      "loss": 0.1408,
+      "loss": 0.0469,
       "step": 9933
     },
     {
       "epoch": 2.76,
       "learning_rate": 1.567863438166806e-06,
-      "loss": 0.0837,
+      "loss": 0.1112,
       "step": 9934
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5660079784766676e-06,
-      "loss": 0.1392,
+      "loss": 0.0192,
       "step": 9935
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5641525187865295e-06,
-      "loss": 0.1989,
+      "loss": 0.0399,
       "step": 9936
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5622970590963912e-06,
-      "loss": 0.1441,
+      "loss": 0.052,
       "step": 9937
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.560441599406253e-06,
-      "loss": 0.082,
+      "loss": 0.0151,
       "step": 9938
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5585861397161147e-06,
-      "loss": 0.0869,
+      "loss": 0.0709,
       "step": 9939
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5567306800259766e-06,
-      "loss": 0.0313,
+      "loss": 0.0903,
       "step": 9940
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5548752203358382e-06,
-      "loss": 0.0837,
+      "loss": 0.0159,
       "step": 9941
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5530197606457e-06,
-      "loss": 0.0848,
+      "loss": 0.135,
       "step": 9942
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.551164300955562e-06,
-      "loss": 0.0309,
+      "loss": 0.0144,
       "step": 9943
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5493088412654236e-06,
-      "loss": 0.2466,
+      "loss": 0.0997,
       "step": 9944
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5474533815752855e-06,
-      "loss": 0.0863,
+      "loss": 0.0502,
       "step": 9945
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5455979218851472e-06,
-      "loss": 0.1404,
+      "loss": 0.0248,
       "step": 9946
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.543742462195009e-06,
-      "loss": 0.0847,
+      "loss": 0.0621,
       "step": 9947
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5418870025048707e-06,
-      "loss": 0.1393,
+      "loss": 0.0127,
       "step": 9948
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5400315428147326e-06,
-      "loss": 0.0304,
+      "loss": 0.1489,
       "step": 9949
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5381760831245942e-06,
-      "loss": 0.0849,
+      "loss": 0.0404,
       "step": 9950
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.536320623434456e-06,
-      "loss": 0.0305,
+      "loss": 0.0825,
       "step": 9951
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5344651637443177e-06,
-      "loss": 0.1361,
+      "loss": 0.0787,
       "step": 9952
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5326097040541796e-06,
-      "loss": 0.1401,
+      "loss": 0.0126,
       "step": 9953
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5307542443640413e-06,
-      "loss": 0.0309,
+      "loss": 0.0567,
       "step": 9954
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5288987846739031e-06,
-      "loss": 0.141,
+      "loss": 0.1391,
       "step": 9955
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5270433249837648e-06,
-      "loss": 0.3079,
+      "loss": 0.1405,
       "step": 9956
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5251878652936267e-06,
-      "loss": 0.2954,
+      "loss": 0.0845,
       "step": 9957
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5233324056034885e-06,
-      "loss": 0.1424,
+      "loss": 0.0672,
       "step": 9958
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5214769459133502e-06,
-      "loss": 0.0863,
+      "loss": 0.0714,
       "step": 9959
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.519621486223212e-06,
-      "loss": 0.0848,
+      "loss": 0.052,
       "step": 9960
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5177660265330737e-06,
-      "loss": 0.1924,
+      "loss": 0.0437,
       "step": 9961
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5159105668429356e-06,
-      "loss": 0.0843,
+      "loss": 0.0519,
       "step": 9962
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5140551071527973e-06,
-      "loss": 0.2513,
+      "loss": 0.0146,
       "step": 9963
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5121996474626591e-06,
-      "loss": 0.1942,
+      "loss": 0.1211,
       "step": 9964
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5103441877725208e-06,
-      "loss": 0.133,
+      "loss": 0.0928,
       "step": 9965
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5084887280823827e-06,
-      "loss": 0.1374,
+      "loss": 0.0676,
       "step": 9966
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5066332683922443e-06,
-      "loss": 0.252,
+      "loss": 0.0837,
       "step": 9967
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5047778087021062e-06,
-      "loss": 0.0299,
+      "loss": 0.1452,
       "step": 9968
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5029223490119678e-06,
-      "loss": 0.1401,
+      "loss": 0.0144,
       "step": 9969
     },
     {
       "epoch": 2.77,
       "learning_rate": 1.5010668893218297e-06,
-      "loss": 0.0297,
+      "loss": 0.0461,
       "step": 9970
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4992114296316914e-06,
-      "loss": 0.4186,
+      "loss": 0.0387,
       "step": 9971
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4973559699415532e-06,
-      "loss": 0.1934,
+      "loss": 0.1719,
       "step": 9972
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4955005102514151e-06,
-      "loss": 0.1419,
+      "loss": 0.0415,
       "step": 9973
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4936450505612768e-06,
-      "loss": 0.1965,
+      "loss": 0.0729,
       "step": 9974
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4917895908711386e-06,
-      "loss": 0.0833,
+      "loss": 0.0559,
       "step": 9975
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4899341311810003e-06,
-      "loss": 0.2491,
+      "loss": 0.1997,
       "step": 9976
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4880786714908622e-06,
-      "loss": 0.0852,
+      "loss": 0.01,
       "step": 9977
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4862232118007238e-06,
-      "loss": 0.1405,
+      "loss": 0.0597,
       "step": 9978
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4843677521105857e-06,
-      "loss": 0.1396,
+      "loss": 0.1113,
       "step": 9979
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4825122924204474e-06,
-      "loss": 0.0863,
+      "loss": 0.0563,
       "step": 9980
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4806568327303092e-06,
-      "loss": 0.0844,
+      "loss": 0.0314,
       "step": 9981
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4788013730401707e-06,
-      "loss": 0.0832,
+      "loss": 0.0646,
       "step": 9982
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4769459133500326e-06,
-      "loss": 0.0882,
+      "loss": 0.0526,
       "step": 9983
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4750904536598942e-06,
-      "loss": 0.138,
+      "loss": 0.1253,
       "step": 9984
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.473234993969756e-06,
-      "loss": 0.0303,
+      "loss": 0.0197,
       "step": 9985
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4713795342796177e-06,
-      "loss": 0.0305,
+      "loss": 0.0637,
       "step": 9986
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4695240745894796e-06,
-      "loss": 0.0862,
+      "loss": 0.0434,
       "step": 9987
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4676686148993413e-06,
-      "loss": 0.1955,
+      "loss": 0.0474,
       "step": 9988
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4658131552092031e-06,
-      "loss": 0.2483,
+      "loss": 0.0452,
       "step": 9989
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4639576955190648e-06,
-      "loss": 0.2501,
+      "loss": 0.0417,
       "step": 9990
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4621022358289267e-06,
-      "loss": 0.0304,
+      "loss": 0.0598,
       "step": 9991
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4602467761387883e-06,
-      "loss": 0.3013,
+      "loss": 0.0261,
       "step": 9992
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4583913164486502e-06,
-      "loss": 0.2482,
+      "loss": 0.0977,
       "step": 9993
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4565358567585119e-06,
-      "loss": 0.1359,
+      "loss": 0.0681,
       "step": 9994
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4546803970683737e-06,
-      "loss": 0.2531,
+      "loss": 0.0422,
       "step": 9995
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4528249373782354e-06,
-      "loss": 0.2516,
+      "loss": 0.0189,
       "step": 9996
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4509694776880973e-06,
-      "loss": 0.0843,
+      "loss": 0.1173,
       "step": 9997
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4491140179979591e-06,
-      "loss": 0.0869,
+      "loss": 0.0594,
       "step": 9998
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4472585583078208e-06,
-      "loss": 0.0304,
+      "loss": 0.0178,
       "step": 9999
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4454030986176827e-06,
-      "loss": 0.03,
+      "loss": 0.0197,
       "step": 10000
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4435476389275443e-06,
-      "loss": 0.2392,
+      "loss": 0.176,
       "step": 10001
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4416921792374062e-06,
-      "loss": 0.0305,
+      "loss": 0.0103,
       "step": 10002
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4398367195472678e-06,
-      "loss": 0.1905,
+      "loss": 0.0094,
       "step": 10003
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4379812598571297e-06,
-      "loss": 0.0861,
+      "loss": 0.0126,
       "step": 10004
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4361258001669914e-06,
-      "loss": 0.0857,
+      "loss": 0.0281,
       "step": 10005
     },
     {
       "epoch": 2.78,
       "learning_rate": 1.4342703404768532e-06,
-      "loss": 0.0855,
+      "loss": 0.0154,
       "step": 10006
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.432414880786715e-06,
-      "loss": 0.1935,
+      "loss": 0.0702,
       "step": 10007
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4305594210965768e-06,
-      "loss": 0.0833,
+      "loss": 0.02,
       "step": 10008
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4287039614064384e-06,
-      "loss": 0.2474,
+      "loss": 0.0634,
       "step": 10009
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4268485017163003e-06,
-      "loss": 0.0835,
+      "loss": 0.0131,
       "step": 10010
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.424993042026162e-06,
-      "loss": 0.1415,
+      "loss": 0.0932,
       "step": 10011
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4231375823360238e-06,
-      "loss": 0.0825,
+      "loss": 0.0243,
       "step": 10012
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4212821226458857e-06,
-      "loss": 0.087,
+      "loss": 0.0127,
       "step": 10013
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4194266629557474e-06,
-      "loss": 0.137,
+      "loss": 0.0859,
       "step": 10014
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4175712032656092e-06,
-      "loss": 0.0849,
+      "loss": 0.0525,
       "step": 10015
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4157157435754709e-06,
-      "loss": 0.2438,
+      "loss": 0.0956,
       "step": 10016
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4138602838853328e-06,
-      "loss": 0.087,
+      "loss": 0.0697,
       "step": 10017
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4120048241951944e-06,
-      "loss": 0.139,
+      "loss": 0.083,
       "step": 10018
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4101493645050563e-06,
-      "loss": 0.0304,
+      "loss": 0.009,
       "step": 10019
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.408293904814918e-06,
-      "loss": 0.0829,
+      "loss": 0.0465,
       "step": 10020
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4064384451247798e-06,
-      "loss": 0.3645,
+      "loss": 0.0536,
       "step": 10021
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4045829854346415e-06,
-      "loss": 0.0303,
+      "loss": 0.2392,
       "step": 10022
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.4027275257445034e-06,
-      "loss": 0.0838,
+      "loss": 0.081,
       "step": 10023
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.400872066054365e-06,
-      "loss": 0.1408,
+      "loss": 0.0661,
       "step": 10024
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3990166063642269e-06,
-      "loss": 0.0835,
+      "loss": 0.0972,
       "step": 10025
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3971611466740885e-06,
-      "loss": 0.0852,
+      "loss": 0.1017,
       "step": 10026
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3953056869839504e-06,
-      "loss": 0.1391,
+      "loss": 0.0424,
       "step": 10027
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3934502272938123e-06,
-      "loss": 0.246,
+      "loss": 0.0156,
       "step": 10028
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.391594767603674e-06,
-      "loss": 0.2475,
+      "loss": 0.1177,
       "step": 10029
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3897393079135358e-06,
-      "loss": 0.0306,
+      "loss": 0.0755,
       "step": 10030
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3878838482233975e-06,
-      "loss": 0.0846,
+      "loss": 0.1349,
       "step": 10031
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3860283885332593e-06,
-      "loss": 0.1939,
+      "loss": 0.0613,
       "step": 10032
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.384172928843121e-06,
-      "loss": 0.0868,
+      "loss": 0.0186,
       "step": 10033
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3823174691529829e-06,
-      "loss": 0.3654,
+      "loss": 0.016,
       "step": 10034
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3804620094628445e-06,
-      "loss": 0.0302,
+      "loss": 0.0612,
       "step": 10035
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3786065497727064e-06,
-      "loss": 0.1359,
+      "loss": 0.0481,
       "step": 10036
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.376751090082568e-06,
-      "loss": 0.1931,
+      "loss": 0.0117,
       "step": 10037
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.37489563039243e-06,
-      "loss": 0.0306,
+      "loss": 0.0765,
       "step": 10038
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3730401707022916e-06,
-      "loss": 0.1391,
+      "loss": 0.0294,
       "step": 10039
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3711847110121535e-06,
-      "loss": 0.1947,
+      "loss": 0.0213,
       "step": 10040
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.3693292513220151e-06,
-      "loss": 0.0847,
+      "loss": 0.0081,
       "step": 10041
     },
     {
       "epoch": 2.79,
       "learning_rate": 1.367473791631877e-06,
-      "loss": 0.0855,
+      "loss": 0.1394,
       "step": 10042
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3656183319417389e-06,
-      "loss": 0.0303,
+      "loss": 0.0129,
       "step": 10043
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3637628722516005e-06,
-      "loss": 0.0308,
+      "loss": 0.0237,
       "step": 10044
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3619074125614624e-06,
-      "loss": 0.0856,
+      "loss": 0.0183,
       "step": 10045
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.360051952871324e-06,
-      "loss": 0.0839,
+      "loss": 0.0199,
       "step": 10046
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.358196493181186e-06,
-      "loss": 0.1909,
+      "loss": 0.0182,
       "step": 10047
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3563410334910476e-06,
-      "loss": 0.3009,
+      "loss": 0.0733,
       "step": 10048
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3544855738009094e-06,
-      "loss": 0.0824,
+      "loss": 0.0875,
       "step": 10049
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.352630114110771e-06,
-      "loss": 0.1388,
+      "loss": 0.1599,
       "step": 10050
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.350774654420633e-06,
-      "loss": 0.0302,
+      "loss": 0.1017,
       "step": 10051
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3489191947304946e-06,
-      "loss": 0.193,
+      "loss": 0.1256,
       "step": 10052
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3470637350403565e-06,
-      "loss": 0.0301,
+      "loss": 0.109,
       "step": 10053
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3452082753502182e-06,
-      "loss": 0.0869,
+      "loss": 0.0536,
       "step": 10054
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.34335281566008e-06,
-      "loss": 0.1949,
+      "loss": 0.0412,
       "step": 10055
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3414973559699417e-06,
-      "loss": 0.0828,
+      "loss": 0.0136,
       "step": 10056
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3396418962798036e-06,
-      "loss": 0.1397,
+      "loss": 0.1214,
       "step": 10057
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.337786436589665e-06,
-      "loss": 0.3047,
+      "loss": 0.1372,
       "step": 10058
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3359309768995269e-06,
-      "loss": 0.2463,
+      "loss": 0.0471,
       "step": 10059
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3340755172093885e-06,
-      "loss": 0.0882,
+      "loss": 0.0118,
       "step": 10060
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3322200575192504e-06,
-      "loss": 0.0307,
+      "loss": 0.0531,
       "step": 10061
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.330364597829112e-06,
-      "loss": 0.0847,
+      "loss": 0.1311,
       "step": 10062
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.328509138138974e-06,
-      "loss": 0.193,
+      "loss": 0.0639,
       "step": 10063
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3266536784488356e-06,
-      "loss": 0.2493,
+      "loss": 0.0217,
       "step": 10064
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3247982187586975e-06,
-      "loss": 0.0847,
+      "loss": 0.007,
       "step": 10065
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3229427590685593e-06,
-      "loss": 0.142,
+      "loss": 0.0148,
       "step": 10066
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.321087299378421e-06,
-      "loss": 0.1929,
+      "loss": 0.0591,
       "step": 10067
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3192318396882829e-06,
-      "loss": 0.3033,
+      "loss": 0.1349,
       "step": 10068
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3173763799981445e-06,
-      "loss": 0.0847,
+      "loss": 0.0234,
       "step": 10069
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3155209203080064e-06,
-      "loss": 0.137,
+      "loss": 0.0117,
       "step": 10070
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.313665460617868e-06,
-      "loss": 0.1345,
+      "loss": 0.05,
       "step": 10071
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.31181000092773e-06,
-      "loss": 0.139,
+      "loss": 0.0725,
       "step": 10072
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3099545412375916e-06,
-      "loss": 0.0817,
+      "loss": 0.0454,
       "step": 10073
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3080990815474535e-06,
-      "loss": 0.0835,
+      "loss": 0.1145,
       "step": 10074
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3062436218573151e-06,
-      "loss": 0.1417,
+      "loss": 0.0133,
       "step": 10075
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.304388162167177e-06,
-      "loss": 0.0872,
+      "loss": 0.0128,
       "step": 10076
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3025327024770386e-06,
-      "loss": 0.0838,
+      "loss": 0.0154,
       "step": 10077
     },
     {
       "epoch": 2.8,
       "learning_rate": 1.3006772427869005e-06,
-      "loss": 0.2482,
+      "loss": 0.0119,
       "step": 10078
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2988217830967622e-06,
-      "loss": 0.1909,
+      "loss": 0.0165,
       "step": 10079
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.296966323406624e-06,
-      "loss": 0.2471,
+      "loss": 0.0161,
       "step": 10080
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.295110863716486e-06,
-      "loss": 0.3035,
+      "loss": 0.0723,
       "step": 10081
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2932554040263476e-06,
-      "loss": 0.087,
+      "loss": 0.0401,
       "step": 10082
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2913999443362094e-06,
-      "loss": 0.1967,
+      "loss": 0.0566,
       "step": 10083
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.289544484646071e-06,
-      "loss": 0.1379,
+      "loss": 0.0443,
       "step": 10084
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.287689024955933e-06,
-      "loss": 0.1376,
+      "loss": 0.1233,
       "step": 10085
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2858335652657946e-06,
-      "loss": 0.0313,
+      "loss": 0.1466,
       "step": 10086
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2839781055756565e-06,
-      "loss": 0.1903,
+      "loss": 0.0661,
       "step": 10087
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2821226458855182e-06,
-      "loss": 0.1371,
+      "loss": 0.0596,
       "step": 10088
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.28026718619538e-06,
-      "loss": 0.2979,
+      "loss": 0.134,
       "step": 10089
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2784117265052417e-06,
-      "loss": 0.194,
+      "loss": 0.0126,
       "step": 10090
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2765562668151036e-06,
-      "loss": 0.0833,
+      "loss": 0.2342,
       "step": 10091
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2747008071249652e-06,
-      "loss": 0.1367,
+      "loss": 0.0817,
       "step": 10092
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.272845347434827e-06,
-      "loss": 0.1898,
+      "loss": 0.1469,
       "step": 10093
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2709898877446888e-06,
-      "loss": 0.2437,
+      "loss": 0.0124,
       "step": 10094
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2691344280545506e-06,
-      "loss": 0.1384,
+      "loss": 0.038,
       "step": 10095
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2672789683644125e-06,
-      "loss": 0.1396,
+      "loss": 0.0462,
       "step": 10096
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2654235086742742e-06,
-      "loss": 0.0843,
+      "loss": 0.0486,
       "step": 10097
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.263568048984136e-06,
-      "loss": 0.1958,
+      "loss": 0.0346,
       "step": 10098
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2617125892939977e-06,
-      "loss": 0.2449,
+      "loss": 0.089,
       "step": 10099
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2598571296038596e-06,
-      "loss": 0.2469,
+      "loss": 0.0118,
       "step": 10100
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2580016699137212e-06,
-      "loss": 0.248,
+      "loss": 0.072,
       "step": 10101
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.256146210223583e-06,
-      "loss": 0.0311,
+      "loss": 0.0527,
       "step": 10102
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2542907505334447e-06,
-      "loss": 0.086,
+      "loss": 0.0633,
       "step": 10103
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2524352908433066e-06,
-      "loss": 0.0824,
+      "loss": 0.0733,
       "step": 10104
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2505798311531683e-06,
-      "loss": 0.0309,
+      "loss": 0.0656,
       "step": 10105
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2487243714630301e-06,
-      "loss": 0.0317,
+      "loss": 0.064,
       "step": 10106
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2468689117728918e-06,
-      "loss": 0.139,
+      "loss": 0.1262,
       "step": 10107
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2450134520827537e-06,
-      "loss": 0.0863,
+      "loss": 0.0548,
       "step": 10108
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2431579923926153e-06,
-      "loss": 0.0842,
+      "loss": 0.0635,
       "step": 10109
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2413025327024772e-06,
-      "loss": 0.2417,
+      "loss": 0.0645,
       "step": 10110
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.239447073012339e-06,
-      "loss": 0.0842,
+      "loss": 0.079,
       "step": 10111
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2375916133222007e-06,
-      "loss": 0.1416,
+      "loss": 0.0296,
       "step": 10112
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2357361536320626e-06,
-      "loss": 0.0869,
+      "loss": 0.1153,
       "step": 10113
     },
     {
       "epoch": 2.81,
       "learning_rate": 1.2338806939419243e-06,
-      "loss": 0.0854,
+      "loss": 0.0188,
       "step": 10114
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.232025234251786e-06,
-      "loss": 0.081,
+      "loss": 0.1452,
       "step": 10115
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.2301697745616478e-06,
-      "loss": 0.0323,
+      "loss": 0.007,
       "step": 10116
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.2283143148715094e-06,
-      "loss": 0.0833,
+      "loss": 0.0435,
       "step": 10117
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.2264588551813713e-06,
-      "loss": 0.1366,
+      "loss": 0.1119,
       "step": 10118
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.224603395491233e-06,
-      "loss": 0.1415,
+      "loss": 0.0096,
       "step": 10119
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.2227479358010948e-06,
-      "loss": 0.0852,
+      "loss": 0.2002,
       "step": 10120
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.2208924761109565e-06,
-      "loss": 0.031,
+      "loss": 0.2265,
       "step": 10121
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.2190370164208184e-06,
-      "loss": 0.1363,
+      "loss": 0.0152,
       "step": 10122
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.21718155673068e-06,
-      "loss": 0.1935,
+      "loss": 0.0141,
       "step": 10123
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.215326097040542e-06,
-      "loss": 0.2482,
+      "loss": 0.0653,
       "step": 10124
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.2134706373504036e-06,
-      "loss": 0.0829,
+      "loss": 0.0067,
       "step": 10125
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.2116151776602654e-06,
-      "loss": 0.1373,
+      "loss": 0.0383,
       "step": 10126
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.209759717970127e-06,
-      "loss": 0.0842,
+      "loss": 0.1311,
       "step": 10127
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.207904258279989e-06,
-      "loss": 0.1388,
+      "loss": 0.1112,
       "step": 10128
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.2060487985898506e-06,
-      "loss": 0.1393,
+      "loss": 0.0893,
       "step": 10129
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.2041933388997125e-06,
-      "loss": 0.0878,
+      "loss": 0.0514,
       "step": 10130
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.2023378792095744e-06,
-      "loss": 0.1381,
+      "loss": 0.0466,
       "step": 10131
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.200482419519436e-06,
-      "loss": 0.0855,
+      "loss": 0.0235,
       "step": 10132
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.1986269598292979e-06,
-      "loss": 0.1913,
+      "loss": 0.0307,
       "step": 10133
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.1967715001391595e-06,
-      "loss": 0.1375,
+      "loss": 0.096,
       "step": 10134
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.1949160404490214e-06,
-      "loss": 0.1409,
+      "loss": 0.0309,
       "step": 10135
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.193060580758883e-06,
-      "loss": 0.2469,
+      "loss": 0.0385,
       "step": 10136
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.191205121068745e-06,
-      "loss": 0.0306,
+      "loss": 0.0703,
       "step": 10137
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.1893496613786066e-06,
-      "loss": 0.0837,
+      "loss": 0.0179,
       "step": 10138
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.1874942016884685e-06,
-      "loss": 0.4146,
+      "loss": 0.1105,
       "step": 10139
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.1856387419983301e-06,
-      "loss": 0.1386,
+      "loss": 0.1145,
       "step": 10140
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.183783282308192e-06,
-      "loss": 0.0862,
+      "loss": 0.0119,
       "step": 10141
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.1819278226180537e-06,
-      "loss": 0.142,
+      "loss": 0.0228,
       "step": 10142
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.1800723629279155e-06,
-      "loss": 0.136,
+      "loss": 0.1764,
       "step": 10143
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.1782169032377772e-06,
-      "loss": 0.139,
+      "loss": 0.0172,
       "step": 10144
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.176361443547639e-06,
-      "loss": 0.0864,
+      "loss": 0.019,
       "step": 10145
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.174505983857501e-06,
-      "loss": 0.1954,
+      "loss": 0.0276,
       "step": 10146
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.1726505241673626e-06,
-      "loss": 0.1411,
+      "loss": 0.2058,
       "step": 10147
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.1707950644772245e-06,
-      "loss": 0.2965,
+      "loss": 0.1773,
       "step": 10148
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.1689396047870861e-06,
-      "loss": 0.1407,
+      "loss": 0.0109,
       "step": 10149
     },
     {
       "epoch": 2.82,
       "learning_rate": 1.167084145096948e-06,
-      "loss": 0.0877,
+      "loss": 0.1071,
       "step": 10150
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1652286854068097e-06,
-      "loss": 0.1418,
+      "loss": 0.0965,
       "step": 10151
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1633732257166715e-06,
-      "loss": 0.1394,
+      "loss": 0.0236,
       "step": 10152
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1615177660265332e-06,
-      "loss": 0.0312,
+      "loss": 0.0915,
       "step": 10153
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1596623063363948e-06,
-      "loss": 0.247,
+      "loss": 0.0177,
       "step": 10154
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1578068466462567e-06,
-      "loss": 0.0857,
+      "loss": 0.0688,
       "step": 10155
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1559513869561184e-06,
-      "loss": 0.1902,
+      "loss": 0.041,
       "step": 10156
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1540959272659802e-06,
-      "loss": 0.1991,
+      "loss": 0.0123,
       "step": 10157
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.152240467575842e-06,
-      "loss": 0.1414,
+      "loss": 0.0138,
       "step": 10158
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1503850078857038e-06,
-      "loss": 0.198,
+      "loss": 0.0216,
       "step": 10159
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1485295481955654e-06,
-      "loss": 0.0315,
+      "loss": 0.0764,
       "step": 10160
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1466740885054273e-06,
-      "loss": 0.1371,
+      "loss": 0.0942,
       "step": 10161
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.144818628815289e-06,
-      "loss": 0.1358,
+      "loss": 0.0119,
       "step": 10162
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1429631691251508e-06,
-      "loss": 0.0837,
+      "loss": 0.0151,
       "step": 10163
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1411077094350125e-06,
-      "loss": 0.3,
+      "loss": 0.0285,
       "step": 10164
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1392522497448744e-06,
-      "loss": 0.1392,
+      "loss": 0.0602,
       "step": 10165
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1373967900547362e-06,
-      "loss": 0.0313,
+      "loss": 0.0079,
       "step": 10166
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1355413303645979e-06,
-      "loss": 0.0307,
+      "loss": 0.0771,
       "step": 10167
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1336858706744598e-06,
-      "loss": 0.0863,
+      "loss": 0.0744,
       "step": 10168
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1318304109843214e-06,
-      "loss": 0.0842,
+      "loss": 0.0777,
       "step": 10169
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1299749512941833e-06,
-      "loss": 0.0867,
+      "loss": 0.0925,
       "step": 10170
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.128119491604045e-06,
-      "loss": 0.1429,
+      "loss": 0.0502,
       "step": 10171
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1262640319139068e-06,
-      "loss": 0.1915,
+      "loss": 0.0586,
       "step": 10172
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1244085722237685e-06,
-      "loss": 0.1373,
+      "loss": 0.0705,
       "step": 10173
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1225531125336303e-06,
-      "loss": 0.0308,
+      "loss": 0.0205,
       "step": 10174
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.120697652843492e-06,
-      "loss": 0.1369,
+      "loss": 0.0234,
       "step": 10175
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1188421931533539e-06,
-      "loss": 0.0817,
+      "loss": 0.0406,
       "step": 10176
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1169867334632155e-06,
-      "loss": 0.3491,
+      "loss": 0.1139,
       "step": 10177
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1151312737730774e-06,
-      "loss": 0.0859,
+      "loss": 0.0812,
       "step": 10178
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.113275814082939e-06,
-      "loss": 0.0838,
+      "loss": 0.0845,
       "step": 10179
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.111420354392801e-06,
-      "loss": 0.1372,
+      "loss": 0.0442,
       "step": 10180
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1095648947026628e-06,
-      "loss": 0.1971,
+      "loss": 0.1057,
       "step": 10181
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1077094350125245e-06,
-      "loss": 0.0858,
+      "loss": 0.1349,
       "step": 10182
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1058539753223863e-06,
-      "loss": 0.0841,
+      "loss": 0.0832,
       "step": 10183
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.103998515632248e-06,
-      "loss": 0.0855,
+      "loss": 0.0736,
       "step": 10184
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1021430559421099e-06,
-      "loss": 0.0318,
+      "loss": 0.0438,
       "step": 10185
     },
     {
       "epoch": 2.83,
       "learning_rate": 1.1002875962519715e-06,
-      "loss": 0.0837,
+      "loss": 0.1931,
       "step": 10186
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0984321365618334e-06,
-      "loss": 0.1895,
+      "loss": 0.1462,
       "step": 10187
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.096576676871695e-06,
-      "loss": 0.1942,
+      "loss": 0.0125,
       "step": 10188
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.094721217181557e-06,
-      "loss": 0.139,
+      "loss": 0.0745,
       "step": 10189
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0928657574914186e-06,
-      "loss": 0.1369,
+      "loss": 0.0167,
       "step": 10190
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0910102978012802e-06,
-      "loss": 0.0308,
+      "loss": 0.0126,
       "step": 10191
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0891548381111421e-06,
-      "loss": 0.2512,
+      "loss": 0.0076,
       "step": 10192
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0872993784210038e-06,
-      "loss": 0.1381,
+      "loss": 0.0634,
       "step": 10193
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0854439187308656e-06,
-      "loss": 0.2486,
+      "loss": 0.1168,
       "step": 10194
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0835884590407273e-06,
-      "loss": 0.0852,
+      "loss": 0.0275,
       "step": 10195
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0817329993505892e-06,
-      "loss": 0.0848,
+      "loss": 0.0595,
       "step": 10196
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0798775396604508e-06,
-      "loss": 0.1938,
+      "loss": 0.1057,
       "step": 10197
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0780220799703127e-06,
-      "loss": 0.1382,
+      "loss": 0.089,
       "step": 10198
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0761666202801744e-06,
-      "loss": 0.1956,
+      "loss": 0.0614,
       "step": 10199
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0743111605900362e-06,
-      "loss": 0.0854,
+      "loss": 0.0718,
       "step": 10200
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.072455700899898e-06,
-      "loss": 0.1957,
+      "loss": 0.0147,
       "step": 10201
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0706002412097598e-06,
-      "loss": 0.193,
+      "loss": 0.0104,
       "step": 10202
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0687447815196216e-06,
-      "loss": 0.1409,
+      "loss": 0.0496,
       "step": 10203
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0668893218294833e-06,
-      "loss": 0.0869,
+      "loss": 0.1189,
       "step": 10204
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0650338621393452e-06,
-      "loss": 0.0865,
+      "loss": 0.0313,
       "step": 10205
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0631784024492068e-06,
-      "loss": 0.1396,
+      "loss": 0.0137,
       "step": 10206
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0613229427590687e-06,
-      "loss": 0.2507,
+      "loss": 0.0901,
       "step": 10207
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0594674830689303e-06,
-      "loss": 0.1418,
+      "loss": 0.0933,
       "step": 10208
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0576120233787922e-06,
-      "loss": 0.2513,
+      "loss": 0.0368,
       "step": 10209
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0557565636886539e-06,
-      "loss": 0.1928,
+      "loss": 0.0154,
       "step": 10210
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0539011039985157e-06,
-      "loss": 0.0864,
+      "loss": 0.0175,
       "step": 10211
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0520456443083774e-06,
-      "loss": 0.136,
+      "loss": 0.1067,
       "step": 10212
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0501901846182393e-06,
-      "loss": 0.1949,
+      "loss": 0.1388,
       "step": 10213
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.048334724928101e-06,
-      "loss": 0.1392,
+      "loss": 0.0912,
       "step": 10214
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0464792652379628e-06,
-      "loss": 0.0304,
+      "loss": 0.0163,
       "step": 10215
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0446238055478247e-06,
-      "loss": 0.0834,
+      "loss": 0.0764,
       "step": 10216
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0427683458576863e-06,
-      "loss": 0.0859,
+      "loss": 0.0504,
       "step": 10217
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0409128861675482e-06,
-      "loss": 0.0863,
+      "loss": 0.0559,
       "step": 10218
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0390574264774099e-06,
-      "loss": 0.2495,
+      "loss": 0.0201,
       "step": 10219
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0372019667872717e-06,
-      "loss": 0.1915,
+      "loss": 0.1127,
       "step": 10220
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0353465070971334e-06,
-      "loss": 0.1383,
+      "loss": 0.1033,
       "step": 10221
     },
     {
       "epoch": 2.84,
       "learning_rate": 1.0334910474069953e-06,
-      "loss": 0.0848,
+      "loss": 0.1505,
       "step": 10222
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.031635587716857e-06,
-      "loss": 0.0854,
+      "loss": 0.0248,
       "step": 10223
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.0297801280267188e-06,
-      "loss": 0.3532,
+      "loss": 0.0586,
       "step": 10224
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.0279246683365805e-06,
-      "loss": 0.2465,
+      "loss": 0.1641,
       "step": 10225
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.0260692086464423e-06,
-      "loss": 0.2451,
+      "loss": 0.0619,
       "step": 10226
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.024213748956304e-06,
-      "loss": 0.2442,
+      "loss": 0.0163,
       "step": 10227
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.0223582892661659e-06,
-      "loss": 0.031,
+      "loss": 0.0134,
       "step": 10228
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.0205028295760275e-06,
-      "loss": 0.1379,
+      "loss": 0.0777,
       "step": 10229
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.0186473698858892e-06,
-      "loss": 0.192,
+      "loss": 0.0685,
       "step": 10230
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.016791910195751e-06,
-      "loss": 0.2479,
+      "loss": 0.0802,
       "step": 10231
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.0149364505056127e-06,
-      "loss": 0.0838,
+      "loss": 0.0182,
       "step": 10232
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.0130809908154746e-06,
-      "loss": 0.3001,
+      "loss": 0.0917,
       "step": 10233
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.0112255311253362e-06,
-      "loss": 0.1976,
+      "loss": 0.0262,
       "step": 10234
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.009370071435198e-06,
-      "loss": 0.1887,
+      "loss": 0.0503,
       "step": 10235
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.00751461174506e-06,
-      "loss": 0.1942,
+      "loss": 0.0996,
       "step": 10236
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.0056591520549216e-06,
-      "loss": 0.14,
+      "loss": 0.0267,
       "step": 10237
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.0038036923647835e-06,
-      "loss": 0.0847,
+      "loss": 0.0316,
       "step": 10238
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.0019482326746452e-06,
-      "loss": 0.0317,
+      "loss": 0.0174,
       "step": 10239
     },
     {
       "epoch": 2.85,
       "learning_rate": 1.000092772984507e-06,
-      "loss": 0.0855,
+      "loss": 0.0305,
       "step": 10240
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.982373132943687e-07,
-      "loss": 0.1928,
+      "loss": 0.0195,
       "step": 10241
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.963818536042306e-07,
-      "loss": 0.1401,
+      "loss": 0.0873,
       "step": 10242
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.945263939140922e-07,
-      "loss": 0.2448,
+      "loss": 0.0549,
       "step": 10243
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.92670934223954e-07,
-      "loss": 0.0804,
+      "loss": 0.0614,
       "step": 10244
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.908154745338157e-07,
-      "loss": 0.1393,
+      "loss": 0.012,
       "step": 10245
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.889600148436776e-07,
-      "loss": 0.4088,
+      "loss": 0.0098,
       "step": 10246
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.871045551535393e-07,
-      "loss": 0.0862,
+      "loss": 0.0132,
       "step": 10247
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.852490954634011e-07,
-      "loss": 0.1915,
+      "loss": 0.0207,
       "step": 10248
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.833936357732628e-07,
-      "loss": 0.032,
+      "loss": 0.2286,
       "step": 10249
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.815381760831247e-07,
-      "loss": 0.0837,
+      "loss": 0.0221,
       "step": 10250
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.796827163929865e-07,
-      "loss": 0.1338,
+      "loss": 0.0543,
       "step": 10251
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.778272567028482e-07,
-      "loss": 0.0854,
+      "loss": 0.071,
       "step": 10252
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.7597179701271e-07,
-      "loss": 0.1922,
+      "loss": 0.0552,
       "step": 10253
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.741163373225717e-07,
-      "loss": 0.1378,
+      "loss": 0.1421,
       "step": 10254
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.722608776324336e-07,
-      "loss": 0.0862,
+      "loss": 0.0765,
       "step": 10255
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.704054179422953e-07,
-      "loss": 0.1932,
+      "loss": 0.0708,
       "step": 10256
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.685499582521571e-07,
-      "loss": 0.1427,
+      "loss": 0.0878,
       "step": 10257
     },
     {
       "epoch": 2.85,
       "learning_rate": 9.666944985620188e-07,
-      "loss": 0.1903,
+      "loss": 0.1051,
       "step": 10258
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.648390388718807e-07,
-      "loss": 0.1923,
+      "loss": 0.0247,
       "step": 10259
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.629835791817423e-07,
-      "loss": 0.0842,
+      "loss": 0.0219,
       "step": 10260
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.611281194916042e-07,
-      "loss": 0.1383,
+      "loss": 0.104,
       "step": 10261
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.592726598014659e-07,
-      "loss": 0.1916,
+      "loss": 0.144,
       "step": 10262
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.574172001113277e-07,
-      "loss": 0.2976,
+      "loss": 0.0461,
       "step": 10263
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.555617404211894e-07,
-      "loss": 0.2509,
+      "loss": 0.0773,
       "step": 10264
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.537062807310513e-07,
-      "loss": 0.1898,
+      "loss": 0.0398,
       "step": 10265
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.51850821040913e-07,
-      "loss": 0.1377,
+      "loss": 0.0439,
       "step": 10266
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.499953613507747e-07,
-      "loss": 0.1947,
+      "loss": 0.1123,
       "step": 10267
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.481399016606364e-07,
-      "loss": 0.0859,
+      "loss": 0.0161,
       "step": 10268
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.462844419704982e-07,
-      "loss": 0.032,
+      "loss": 0.019,
       "step": 10269
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.4442898228036e-07,
-      "loss": 0.084,
+      "loss": 0.0116,
       "step": 10270
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.425735225902217e-07,
-      "loss": 0.0873,
+      "loss": 0.02,
       "step": 10271
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.407180629000835e-07,
-      "loss": 0.1862,
+      "loss": 0.0577,
       "step": 10272
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.388626032099453e-07,
-      "loss": 0.1421,
+      "loss": 0.0641,
       "step": 10273
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.37007143519807e-07,
-      "loss": 0.1949,
+      "loss": 0.0153,
       "step": 10274
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.351516838296688e-07,
-      "loss": 0.1399,
+      "loss": 0.0203,
       "step": 10275
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.332962241395306e-07,
-      "loss": 0.1927,
+      "loss": 0.0218,
       "step": 10276
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.314407644493924e-07,
-      "loss": 0.2455,
+      "loss": 0.0448,
       "step": 10277
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.295853047592542e-07,
-      "loss": 0.0321,
+      "loss": 0.1394,
       "step": 10278
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.27729845069116e-07,
-      "loss": 0.0865,
+      "loss": 0.0256,
       "step": 10279
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.258743853789777e-07,
-      "loss": 0.0325,
+      "loss": 0.113,
       "step": 10280
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.240189256888395e-07,
-      "loss": 0.199,
+      "loss": 0.0993,
       "step": 10281
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.221634659987013e-07,
-      "loss": 0.0832,
+      "loss": 0.0539,
       "step": 10282
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.20308006308563e-07,
-      "loss": 0.351,
+      "loss": 0.0938,
       "step": 10283
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.184525466184248e-07,
-      "loss": 0.1434,
+      "loss": 0.1164,
       "step": 10284
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.165970869282865e-07,
-      "loss": 0.086,
+      "loss": 0.0607,
       "step": 10285
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.147416272381483e-07,
-      "loss": 0.1388,
+      "loss": 0.0171,
       "step": 10286
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.128861675480101e-07,
-      "loss": 0.1938,
+      "loss": 0.0438,
       "step": 10287
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.110307078578718e-07,
-      "loss": 0.0326,
+      "loss": 0.0953,
       "step": 10288
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.091752481677336e-07,
-      "loss": 0.139,
+      "loss": 0.0118,
       "step": 10289
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.073197884775954e-07,
-      "loss": 0.1387,
+      "loss": 0.0171,
       "step": 10290
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.054643287874571e-07,
-      "loss": 0.1378,
+      "loss": 0.1301,
       "step": 10291
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.03608869097319e-07,
-      "loss": 0.0871,
+      "loss": 0.0922,
       "step": 10292
     },
     {
       "epoch": 2.86,
       "learning_rate": 9.017534094071808e-07,
-      "loss": 0.1934,
+      "loss": 0.084,
       "step": 10293
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.998979497170425e-07,
-      "loss": 0.14,
+      "loss": 0.093,
       "step": 10294
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.980424900269043e-07,
-      "loss": 0.1892,
+      "loss": 0.0536,
       "step": 10295
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.961870303367661e-07,
-      "loss": 0.0872,
+      "loss": 0.0086,
       "step": 10296
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.943315706466278e-07,
-      "loss": 0.1398,
+      "loss": 0.0361,
       "step": 10297
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.924761109564896e-07,
-      "loss": 0.086,
+      "loss": 0.0603,
       "step": 10298
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.906206512663514e-07,
-      "loss": 0.0334,
+      "loss": 0.047,
       "step": 10299
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.887651915762131e-07,
-      "loss": 0.1388,
+      "loss": 0.1052,
       "step": 10300
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.869097318860749e-07,
-      "loss": 0.0839,
+      "loss": 0.0106,
       "step": 10301
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.850542721959367e-07,
-      "loss": 0.0878,
+      "loss": 0.0479,
       "step": 10302
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.831988125057984e-07,
-      "loss": 0.2445,
+      "loss": 0.1083,
       "step": 10303
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.813433528156602e-07,
-      "loss": 0.0867,
+      "loss": 0.099,
       "step": 10304
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.794878931255218e-07,
-      "loss": 0.301,
+      "loss": 0.0439,
       "step": 10305
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.776324334353836e-07,
-      "loss": 0.0855,
+      "loss": 0.0126,
       "step": 10306
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.757769737452454e-07,
-      "loss": 0.0864,
+      "loss": 0.0618,
       "step": 10307
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.739215140551071e-07,
-      "loss": 0.1396,
+      "loss": 0.0617,
       "step": 10308
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.720660543649689e-07,
-      "loss": 0.1379,
+      "loss": 0.0591,
       "step": 10309
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.702105946748307e-07,
-      "loss": 0.3532,
+      "loss": 0.0871,
       "step": 10310
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.683551349846925e-07,
-      "loss": 0.2445,
+      "loss": 0.1141,
       "step": 10311
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.664996752945543e-07,
-      "loss": 0.0863,
+      "loss": 0.1059,
       "step": 10312
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.646442156044161e-07,
-      "loss": 0.1395,
+      "loss": 0.2237,
       "step": 10313
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.627887559142778e-07,
-      "loss": 0.0846,
+      "loss": 0.0405,
       "step": 10314
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.609332962241396e-07,
-      "loss": 0.0327,
+      "loss": 0.047,
       "step": 10315
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.590778365340014e-07,
-      "loss": 0.0875,
+      "loss": 0.016,
       "step": 10316
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.572223768438631e-07,
-      "loss": 0.1366,
+      "loss": 0.0807,
       "step": 10317
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.553669171537249e-07,
-      "loss": 0.1902,
+      "loss": 0.0164,
       "step": 10318
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.535114574635867e-07,
-      "loss": 0.1934,
+      "loss": 0.1319,
       "step": 10319
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.516559977734484e-07,
-      "loss": 0.1358,
+      "loss": 0.0774,
       "step": 10320
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.498005380833102e-07,
-      "loss": 0.0838,
+      "loss": 0.0818,
       "step": 10321
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.47945078393172e-07,
-      "loss": 0.0865,
+      "loss": 0.0234,
       "step": 10322
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.460896187030337e-07,
-      "loss": 0.1396,
+      "loss": 0.0472,
       "step": 10323
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.442341590128955e-07,
-      "loss": 0.2492,
+      "loss": 0.0633,
       "step": 10324
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.423786993227572e-07,
-      "loss": 0.087,
+      "loss": 0.0097,
       "step": 10325
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.405232396326191e-07,
-      "loss": 0.1389,
+      "loss": 0.0221,
       "step": 10326
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.386677799424809e-07,
-      "loss": 0.0883,
+      "loss": 0.069,
       "step": 10327
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.368123202523426e-07,
-      "loss": 0.0317,
+      "loss": 0.0135,
       "step": 10328
     },
     {
       "epoch": 2.87,
       "learning_rate": 8.349568605622044e-07,
-      "loss": 0.0878,
+      "loss": 0.0164,
       "step": 10329
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.331014008720662e-07,
-      "loss": 0.1924,
+      "loss": 0.0206,
       "step": 10330
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.312459411819279e-07,
-      "loss": 0.1393,
+      "loss": 0.0302,
       "step": 10331
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.293904814917897e-07,
-      "loss": 0.1399,
+      "loss": 0.1034,
       "step": 10332
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.275350218016515e-07,
-      "loss": 0.0317,
+      "loss": 0.1437,
       "step": 10333
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.256795621115132e-07,
-      "loss": 0.1351,
+      "loss": 0.0141,
       "step": 10334
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.23824102421375e-07,
-      "loss": 0.0871,
+      "loss": 0.0817,
       "step": 10335
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.219686427312368e-07,
-      "loss": 0.1922,
+      "loss": 0.0824,
       "step": 10336
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.201131830410985e-07,
-      "loss": 0.1395,
+      "loss": 0.0491,
       "step": 10337
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.182577233509603e-07,
-      "loss": 0.1857,
+      "loss": 0.1059,
       "step": 10338
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.16402263660822e-07,
-      "loss": 0.1434,
+      "loss": 0.0945,
       "step": 10339
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.145468039706838e-07,
-      "loss": 0.1898,
+      "loss": 0.0509,
       "step": 10340
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.126913442805457e-07,
-      "loss": 0.0863,
+      "loss": 0.1001,
       "step": 10341
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.108358845904075e-07,
-      "loss": 0.0327,
+      "loss": 0.0441,
       "step": 10342
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.08980424900269e-07,
-      "loss": 0.0321,
+      "loss": 0.019,
       "step": 10343
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.071249652101308e-07,
-      "loss": 0.2463,
+      "loss": 0.0513,
       "step": 10344
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.052695055199925e-07,
-      "loss": 0.0858,
+      "loss": 0.057,
       "step": 10345
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.034140458298544e-07,
-      "loss": 0.0881,
+      "loss": 0.0429,
       "step": 10346
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.015585861397162e-07,
-      "loss": 0.2401,
+      "loss": 0.0409,
       "step": 10347
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.997031264495779e-07,
-      "loss": 0.0861,
+      "loss": 0.1348,
       "step": 10348
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.978476667594397e-07,
-      "loss": 0.3017,
+      "loss": 0.0724,
       "step": 10349
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.959922070693015e-07,
-      "loss": 0.0901,
+      "loss": 0.0159,
       "step": 10350
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.941367473791632e-07,
-      "loss": 0.1364,
+      "loss": 0.0676,
       "step": 10351
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.92281287689025e-07,
-      "loss": 0.1936,
+      "loss": 0.0164,
       "step": 10352
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.904258279988868e-07,
-      "loss": 0.0326,
+      "loss": 0.0421,
       "step": 10353
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.885703683087485e-07,
-      "loss": 0.1949,
+      "loss": 0.0309,
       "step": 10354
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.867149086186103e-07,
-      "loss": 0.0317,
+      "loss": 0.0103,
       "step": 10355
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.84859448928472e-07,
-      "loss": 0.2424,
+      "loss": 0.0185,
       "step": 10356
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.830039892383338e-07,
-      "loss": 0.0853,
+      "loss": 0.0065,
       "step": 10357
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.811485295481956e-07,
-      "loss": 0.139,
+      "loss": 0.1519,
       "step": 10358
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.792930698580573e-07,
-      "loss": 0.0832,
+      "loss": 0.201,
       "step": 10359
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.774376101679191e-07,
-      "loss": 0.0321,
+      "loss": 0.1277,
       "step": 10360
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.75582150477781e-07,
-      "loss": 0.1949,
+      "loss": 0.0448,
       "step": 10361
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.737266907876427e-07,
-      "loss": 0.1419,
+      "loss": 0.0871,
       "step": 10362
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.718712310975045e-07,
-      "loss": 0.141,
+      "loss": 0.0597,
       "step": 10363
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.700157714073663e-07,
-      "loss": 0.0853,
+      "loss": 0.059,
       "step": 10364
     },
     {
       "epoch": 2.88,
       "learning_rate": 7.68160311717228e-07,
-      "loss": 0.0854,
+      "loss": 0.0849,
       "step": 10365
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.663048520270898e-07,
-      "loss": 0.0851,
+      "loss": 0.0425,
       "step": 10366
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.644493923369516e-07,
-      "loss": 0.1401,
+      "loss": 0.0409,
       "step": 10367
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.625939326468133e-07,
-      "loss": 0.0856,
+      "loss": 0.005,
       "step": 10368
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.607384729566751e-07,
-      "loss": 0.1929,
+      "loss": 0.0715,
       "step": 10369
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.588830132665369e-07,
-      "loss": 0.0327,
+      "loss": 0.0158,
       "step": 10370
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.570275535763986e-07,
-      "loss": 0.0875,
+      "loss": 0.0197,
       "step": 10371
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.551720938862604e-07,
-      "loss": 0.032,
+      "loss": 0.0692,
       "step": 10372
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.533166341961222e-07,
-      "loss": 0.0319,
+      "loss": 0.0162,
       "step": 10373
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.514611745059839e-07,
-      "loss": 0.1924,
+      "loss": 0.022,
       "step": 10374
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.496057148158457e-07,
-      "loss": 0.0317,
+      "loss": 0.018,
       "step": 10375
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.477502551257076e-07,
-      "loss": 0.1912,
+      "loss": 0.085,
       "step": 10376
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.458947954355693e-07,
-      "loss": 0.1416,
+      "loss": 0.0473,
       "step": 10377
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.440393357454311e-07,
-      "loss": 0.1881,
+      "loss": 0.1538,
       "step": 10378
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.421838760552929e-07,
-      "loss": 0.194,
+      "loss": 0.0678,
       "step": 10379
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.403284163651546e-07,
-      "loss": 0.0868,
+      "loss": 0.1183,
       "step": 10380
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.384729566750163e-07,
-      "loss": 0.1407,
+      "loss": 0.1167,
       "step": 10381
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.36617496984878e-07,
-      "loss": 0.1368,
+      "loss": 0.0422,
       "step": 10382
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.347620372947398e-07,
-      "loss": 0.1935,
+      "loss": 0.1047,
       "step": 10383
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.329065776046016e-07,
-      "loss": 0.14,
+      "loss": 0.0186,
       "step": 10384
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.310511179144633e-07,
-      "loss": 0.0843,
+      "loss": 0.1083,
       "step": 10385
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.291956582243251e-07,
-      "loss": 0.1387,
+      "loss": 0.0342,
       "step": 10386
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.273401985341869e-07,
-      "loss": 0.0845,
+      "loss": 0.0237,
       "step": 10387
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.254847388440486e-07,
-      "loss": 0.0837,
+      "loss": 0.1076,
       "step": 10388
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.236292791539104e-07,
-      "loss": 0.0319,
+      "loss": 0.0817,
       "step": 10389
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.217738194637722e-07,
-      "loss": 0.1403,
+      "loss": 0.0468,
       "step": 10390
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.199183597736339e-07,
-      "loss": 0.1396,
+      "loss": 0.0113,
       "step": 10391
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.180629000834957e-07,
-      "loss": 0.1374,
+      "loss": 0.037,
       "step": 10392
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.162074403933575e-07,
-      "loss": 0.1943,
+      "loss": 0.0467,
       "step": 10393
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.143519807032192e-07,
-      "loss": 0.0832,
+      "loss": 0.0784,
       "step": 10394
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.12496521013081e-07,
-      "loss": 0.1916,
+      "loss": 0.0456,
       "step": 10395
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.106410613229429e-07,
-      "loss": 0.1945,
+      "loss": 0.0215,
       "step": 10396
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.087856016328046e-07,
-      "loss": 0.1405,
+      "loss": 0.0759,
       "step": 10397
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.069301419426664e-07,
-      "loss": 0.1334,
+      "loss": 0.0553,
       "step": 10398
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.050746822525281e-07,
-      "loss": 0.1412,
+      "loss": 0.2002,
       "step": 10399
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.032192225623899e-07,
-      "loss": 0.1366,
+      "loss": 0.0607,
       "step": 10400
     },
     {
       "epoch": 2.89,
       "learning_rate": 7.013637628722517e-07,
-      "loss": 0.1358,
+      "loss": 0.0455,
       "step": 10401
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.995083031821134e-07,
-      "loss": 0.2482,
+      "loss": 0.1021,
       "step": 10402
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.976528434919752e-07,
-      "loss": 0.0872,
+      "loss": 0.0134,
       "step": 10403
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.95797383801837e-07,
-      "loss": 0.0309,
+      "loss": 0.1584,
       "step": 10404
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.939419241116987e-07,
-      "loss": 0.0865,
+      "loss": 0.1306,
       "step": 10405
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.920864644215605e-07,
-      "loss": 0.2515,
+      "loss": 0.1368,
       "step": 10406
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.902310047314223e-07,
-      "loss": 0.1937,
+      "loss": 0.0122,
       "step": 10407
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.88375545041284e-07,
-      "loss": 0.2966,
+      "loss": 0.1537,
       "step": 10408
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.865200853511458e-07,
-      "loss": 0.0311,
+      "loss": 0.0652,
       "step": 10409
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.846646256610076e-07,
-      "loss": 0.1384,
+      "loss": 0.0303,
       "step": 10410
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.828091659708694e-07,
-      "loss": 0.0848,
+      "loss": 0.0156,
       "step": 10411
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.809537062807312e-07,
-      "loss": 0.1953,
+      "loss": 0.1284,
       "step": 10412
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.79098246590593e-07,
-      "loss": 0.0865,
+      "loss": 0.1043,
       "step": 10413
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.772427869004547e-07,
-      "loss": 0.0839,
+      "loss": 0.0102,
       "step": 10414
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.753873272103165e-07,
-      "loss": 0.1379,
+      "loss": 0.0308,
       "step": 10415
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.735318675201783e-07,
-      "loss": 0.0863,
+      "loss": 0.1367,
       "step": 10416
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.7167640783004e-07,
-      "loss": 0.0319,
+      "loss": 0.0583,
       "step": 10417
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.698209481399018e-07,
-      "loss": 0.0863,
+      "loss": 0.0626,
       "step": 10418
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.679654884497634e-07,
-      "loss": 0.1393,
+      "loss": 0.1605,
       "step": 10419
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.661100287596252e-07,
-      "loss": 0.2425,
+      "loss": 0.0247,
       "step": 10420
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.64254569069487e-07,
-      "loss": 0.139,
+      "loss": 0.0889,
       "step": 10421
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.623991093793487e-07,
-      "loss": 0.1387,
+      "loss": 0.0193,
       "step": 10422
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.605436496892105e-07,
-      "loss": 0.1427,
+      "loss": 0.055,
       "step": 10423
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.586881899990723e-07,
-      "loss": 0.1934,
+      "loss": 0.0786,
       "step": 10424
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.56832730308934e-07,
-      "loss": 0.0874,
+      "loss": 0.0707,
       "step": 10425
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.549772706187958e-07,
-      "loss": 0.0868,
+      "loss": 0.0737,
       "step": 10426
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.531218109286576e-07,
-      "loss": 0.1418,
+      "loss": 0.0144,
       "step": 10427
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.512663512385193e-07,
-      "loss": 0.0849,
+      "loss": 0.0544,
       "step": 10428
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.494108915483811e-07,
-      "loss": 0.1378,
+      "loss": 0.1018,
       "step": 10429
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.47555431858243e-07,
-      "loss": 0.1912,
+      "loss": 0.1747,
       "step": 10430
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.456999721681047e-07,
-      "loss": 0.2976,
+      "loss": 0.0827,
       "step": 10431
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.438445124779665e-07,
-      "loss": 0.0897,
+      "loss": 0.0499,
       "step": 10432
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.419890527878283e-07,
-      "loss": 0.0321,
+      "loss": 0.1171,
       "step": 10433
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.4013359309769e-07,
-      "loss": 0.3007,
+      "loss": 0.0162,
       "step": 10434
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.382781334075518e-07,
-      "loss": 0.0865,
+      "loss": 0.0156,
       "step": 10435
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.364226737174135e-07,
-      "loss": 0.2389,
+      "loss": 0.1229,
       "step": 10436
     },
     {
       "epoch": 2.9,
       "learning_rate": 6.345672140272753e-07,
-      "loss": 0.2457,
+      "loss": 0.0487,
       "step": 10437
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.327117543371371e-07,
-      "loss": 0.3002,
+      "loss": 0.0159,
       "step": 10438
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.308562946469988e-07,
-      "loss": 0.1933,
+      "loss": 0.092,
       "step": 10439
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.290008349568606e-07,
-      "loss": 0.2466,
+      "loss": 0.0739,
       "step": 10440
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.271453752667224e-07,
-      "loss": 0.1376,
+      "loss": 0.0236,
       "step": 10441
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.252899155765841e-07,
-      "loss": 0.1921,
+      "loss": 0.0489,
       "step": 10442
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.234344558864459e-07,
-      "loss": 0.1366,
+      "loss": 0.0952,
       "step": 10443
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.215789961963077e-07,
-      "loss": 0.0872,
+      "loss": 0.0502,
       "step": 10444
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.197235365061695e-07,
-      "loss": 0.0316,
+      "loss": 0.0176,
       "step": 10445
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.178680768160313e-07,
-      "loss": 0.0859,
+      "loss": 0.0117,
       "step": 10446
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.16012617125893e-07,
-      "loss": 0.1348,
+      "loss": 0.1065,
       "step": 10447
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.141571574357547e-07,
-      "loss": 0.1391,
+      "loss": 0.0103,
       "step": 10448
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.123016977456165e-07,
-      "loss": 0.0845,
+      "loss": 0.018,
       "step": 10449
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.104462380554783e-07,
-      "loss": 0.1897,
+      "loss": 0.0536,
       "step": 10450
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.0859077836534e-07,
-      "loss": 0.138,
+      "loss": 0.0133,
       "step": 10451
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.067353186752018e-07,
-      "loss": 0.244,
+      "loss": 0.0523,
       "step": 10452
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.048798589850635e-07,
-      "loss": 0.1915,
+      "loss": 0.0836,
       "step": 10453
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.030243992949253e-07,
-      "loss": 0.1377,
+      "loss": 0.0497,
       "step": 10454
     },
     {
       "epoch": 2.91,
       "learning_rate": 6.011689396047872e-07,
-      "loss": 0.0855,
+      "loss": 0.2478,
       "step": 10455
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.993134799146489e-07,
-      "loss": 0.0866,
+      "loss": 0.109,
       "step": 10456
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.974580202245107e-07,
-      "loss": 0.1392,
+      "loss": 0.0104,
       "step": 10457
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.956025605343725e-07,
-      "loss": 0.2936,
+      "loss": 0.1061,
       "step": 10458
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.937471008442342e-07,
-      "loss": 0.0885,
+      "loss": 0.1224,
       "step": 10459
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.91891641154096e-07,
-      "loss": 0.0851,
+      "loss": 0.0515,
       "step": 10460
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.900361814639578e-07,
-      "loss": 0.0862,
+      "loss": 0.0661,
       "step": 10461
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.881807217738195e-07,
-      "loss": 0.0869,
+      "loss": 0.0892,
       "step": 10462
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.863252620836813e-07,
-      "loss": 0.1372,
+      "loss": 0.0333,
       "step": 10463
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.844698023935431e-07,
-      "loss": 0.0313,
+      "loss": 0.048,
       "step": 10464
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.826143427034048e-07,
-      "loss": 0.141,
+      "loss": 0.084,
       "step": 10465
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.807588830132666e-07,
-      "loss": 0.0314,
+      "loss": 0.0157,
       "step": 10466
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.789034233231284e-07,
-      "loss": 0.0846,
+      "loss": 0.0506,
       "step": 10467
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.770479636329901e-07,
-      "loss": 0.1897,
+      "loss": 0.1483,
       "step": 10468
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.751925039428519e-07,
-      "loss": 0.1403,
+      "loss": 0.0149,
       "step": 10469
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.733370442527137e-07,
-      "loss": 0.1357,
+      "loss": 0.0677,
       "step": 10470
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.714815845625754e-07,
-      "loss": 0.1378,
+      "loss": 0.0184,
       "step": 10471
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.696261248724372e-07,
-      "loss": 0.2415,
+      "loss": 0.0943,
       "step": 10472
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.677706651822989e-07,
-      "loss": 0.1874,
+      "loss": 0.0865,
       "step": 10473
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.659152054921607e-07,
-      "loss": 0.1373,
+      "loss": 0.0485,
       "step": 10474
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.640597458020225e-07,
-      "loss": 0.0877,
+      "loss": 0.1826,
       "step": 10475
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.622042861118842e-07,
-      "loss": 0.2479,
+      "loss": 0.0381,
       "step": 10476
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.60348826421746e-07,
-      "loss": 0.1386,
+      "loss": 0.0174,
       "step": 10477
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.584933667316078e-07,
-      "loss": 0.0836,
+      "loss": 0.1203,
       "step": 10478
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.566379070414695e-07,
-      "loss": 0.03,
+      "loss": 0.1108,
       "step": 10479
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.547824473513314e-07,
-      "loss": 0.0865,
+      "loss": 0.0521,
       "step": 10480
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.529269876611932e-07,
-      "loss": 0.0846,
+      "loss": 0.0438,
       "step": 10481
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.510715279710549e-07,
-      "loss": 0.1395,
+      "loss": 0.0628,
       "step": 10482
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.492160682809167e-07,
-      "loss": 0.1372,
+      "loss": 0.1428,
       "step": 10483
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.473606085907785e-07,
-      "loss": 0.1389,
+      "loss": 0.1039,
       "step": 10484
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.455051489006401e-07,
-      "loss": 0.1936,
+      "loss": 0.0143,
       "step": 10485
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.436496892105019e-07,
-      "loss": 0.0848,
+      "loss": 0.0589,
       "step": 10486
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.417942295203637e-07,
-      "loss": 0.192,
+      "loss": 0.1215,
       "step": 10487
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.399387698302254e-07,
-      "loss": 0.0309,
+      "loss": 0.0216,
       "step": 10488
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.380833101400872e-07,
-      "loss": 0.1384,
+      "loss": 0.0098,
       "step": 10489
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.36227850449949e-07,
-      "loss": 0.2471,
+      "loss": 0.0978,
       "step": 10490
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.343723907598108e-07,
-      "loss": 0.0314,
+      "loss": 0.0835,
       "step": 10491
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.325169310696726e-07,
-      "loss": 0.1387,
+      "loss": 0.0082,
       "step": 10492
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.306614713795343e-07,
-      "loss": 0.0851,
+      "loss": 0.1399,
       "step": 10493
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.288060116893961e-07,
-      "loss": 0.1377,
+      "loss": 0.0529,
       "step": 10494
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.269505519992579e-07,
-      "loss": 0.0317,
+      "loss": 0.0498,
       "step": 10495
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.250950923091196e-07,
-      "loss": 0.0315,
+      "loss": 0.0113,
       "step": 10496
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.232396326189814e-07,
-      "loss": 0.0804,
+      "loss": 0.0115,
       "step": 10497
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.213841729288432e-07,
-      "loss": 0.0857,
+      "loss": 0.1537,
       "step": 10498
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.195287132387049e-07,
-      "loss": 0.1923,
+      "loss": 0.0545,
       "step": 10499
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.176732535485667e-07,
-      "loss": 0.0316,
+      "loss": 0.0672,
       "step": 10500
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.158177938584285e-07,
-      "loss": 0.1952,
+      "loss": 0.0232,
       "step": 10501
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.139623341682902e-07,
-      "loss": 0.2473,
+      "loss": 0.1498,
       "step": 10502
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.12106874478152e-07,
-      "loss": 0.0876,
+      "loss": 0.093,
       "step": 10503
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.102514147880138e-07,
-      "loss": 0.1391,
+      "loss": 0.0114,
       "step": 10504
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.083959550978755e-07,
-      "loss": 0.0831,
+      "loss": 0.015,
       "step": 10505
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.065404954077373e-07,
-      "loss": 0.2441,
+      "loss": 0.0746,
       "step": 10506
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.04685035717599e-07,
-      "loss": 0.0852,
+      "loss": 0.0514,
       "step": 10507
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.028295760274608e-07,
-      "loss": 0.1413,
+      "loss": 0.0115,
       "step": 10508
     },
     {
       "epoch": 2.92,
       "learning_rate": 5.009741163373226e-07,
-      "loss": 0.1917,
+      "loss": 0.0776,
       "step": 10509
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.991186566471843e-07,
-      "loss": 0.1354,
+      "loss": 0.0177,
       "step": 10510
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.972631969570461e-07,
-      "loss": 0.0838,
+      "loss": 0.1928,
       "step": 10511
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.954077372669079e-07,
-      "loss": 0.0871,
+      "loss": 0.1063,
       "step": 10512
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.935522775767696e-07,
-      "loss": 0.1425,
+      "loss": 0.1605,
       "step": 10513
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.916968178866314e-07,
-      "loss": 0.1426,
+      "loss": 0.0138,
       "step": 10514
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.898413581964933e-07,
-      "loss": 0.1407,
+      "loss": 0.0393,
       "step": 10515
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.87985898506355e-07,
-      "loss": 0.2461,
+      "loss": 0.0148,
       "step": 10516
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.861304388162168e-07,
-      "loss": 0.1399,
+      "loss": 0.1484,
       "step": 10517
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.842749791260786e-07,
-      "loss": 0.0873,
+      "loss": 0.0219,
       "step": 10518
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.824195194359403e-07,
-      "loss": 0.1404,
+      "loss": 0.012,
       "step": 10519
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.805640597458021e-07,
-      "loss": 0.3016,
+      "loss": 0.1033,
       "step": 10520
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.787086000556639e-07,
-      "loss": 0.2453,
+      "loss": 0.0408,
       "step": 10521
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.768531403655256e-07,
-      "loss": 0.1903,
+      "loss": 0.0209,
       "step": 10522
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.7499768067538734e-07,
-      "loss": 0.1399,
+      "loss": 0.0482,
       "step": 10523
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.731422209852491e-07,
-      "loss": 0.0824,
+      "loss": 0.016,
       "step": 10524
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.7128676129511087e-07,
-      "loss": 0.1399,
+      "loss": 0.0499,
       "step": 10525
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.6943130160497263e-07,
-      "loss": 0.1922,
+      "loss": 0.1465,
       "step": 10526
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.675758419148344e-07,
-      "loss": 0.3046,
+      "loss": 0.0729,
       "step": 10527
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.657203822246962e-07,
-      "loss": 0.0842,
+      "loss": 0.1424,
       "step": 10528
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.63864922534558e-07,
-      "loss": 0.1384,
+      "loss": 0.1382,
       "step": 10529
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.6200946284441974e-07,
-      "loss": 0.085,
+      "loss": 0.0191,
       "step": 10530
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.601540031542815e-07,
-      "loss": 0.1404,
+      "loss": 0.0288,
       "step": 10531
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.582985434641433e-07,
-      "loss": 0.3567,
+      "loss": 0.0214,
       "step": 10532
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.5644308377400504e-07,
-      "loss": 0.0829,
+      "loss": 0.0176,
       "step": 10533
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.545876240838668e-07,
-      "loss": 0.2428,
+      "loss": 0.0705,
       "step": 10534
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.5273216439372857e-07,
-      "loss": 0.0882,
+      "loss": 0.0817,
       "step": 10535
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.508767047035904e-07,
-      "loss": 0.1357,
+      "loss": 0.081,
       "step": 10536
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.4902124501345215e-07,
-      "loss": 0.1397,
+      "loss": 0.0251,
       "step": 10537
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.471657853233139e-07,
-      "loss": 0.1382,
+      "loss": 0.0764,
       "step": 10538
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.453103256331757e-07,
-      "loss": 0.031,
+      "loss": 0.0186,
       "step": 10539
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.4345486594303744e-07,
-      "loss": 0.0322,
+      "loss": 0.0138,
       "step": 10540
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.415994062528992e-07,
-      "loss": 0.1391,
+      "loss": 0.0318,
       "step": 10541
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.397439465627609e-07,
-      "loss": 0.1949,
+      "loss": 0.0229,
       "step": 10542
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.378884868726227e-07,
-      "loss": 0.1377,
+      "loss": 0.0811,
       "step": 10543
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.3603302718248445e-07,
-      "loss": 0.1394,
+      "loss": 0.0874,
       "step": 10544
     },
     {
       "epoch": 2.93,
       "learning_rate": 4.3417756749234627e-07,
-      "loss": 0.1959,
+      "loss": 0.1433,
       "step": 10545
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.3232210780220803e-07,
-      "loss": 0.0857,
+      "loss": 0.0647,
       "step": 10546
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.304666481120698e-07,
-      "loss": 0.1398,
+      "loss": 0.0327,
       "step": 10547
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.2861118842193156e-07,
-      "loss": 0.1365,
+      "loss": 0.0249,
       "step": 10548
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.2675572873179333e-07,
-      "loss": 0.0842,
+      "loss": 0.0229,
       "step": 10549
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.249002690416551e-07,
-      "loss": 0.1905,
+      "loss": 0.0329,
       "step": 10550
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.2304480935151686e-07,
-      "loss": 0.0827,
+      "loss": 0.0515,
       "step": 10551
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.211893496613786e-07,
-      "loss": 0.1941,
+      "loss": 0.0508,
       "step": 10552
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.1933388997124044e-07,
-      "loss": 0.0306,
+      "loss": 0.0243,
       "step": 10553
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.174784302811022e-07,
-      "loss": 0.1426,
+      "loss": 0.0223,
       "step": 10554
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.1562297059096397e-07,
-      "loss": 0.085,
+      "loss": 0.1112,
       "step": 10555
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.1376751090082573e-07,
-      "loss": 0.0318,
+      "loss": 0.0096,
       "step": 10556
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.119120512106875e-07,
-      "loss": 0.192,
+      "loss": 0.0174,
       "step": 10557
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.1005659152054926e-07,
-      "loss": 0.1957,
+      "loss": 0.0203,
       "step": 10558
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.08201131830411e-07,
-      "loss": 0.0822,
+      "loss": 0.0608,
       "step": 10559
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.0634567214027284e-07,
-      "loss": 0.1363,
+      "loss": 0.0247,
       "step": 10560
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.044902124501345e-07,
-      "loss": 0.1978,
+      "loss": 0.1765,
       "step": 10561
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.0263475275999627e-07,
-      "loss": 0.1417,
+      "loss": 0.0181,
       "step": 10562
     },
     {
       "epoch": 2.94,
       "learning_rate": 4.007792930698581e-07,
-      "loss": 0.2477,
+      "loss": 0.0493,
       "step": 10563
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.9892383337971985e-07,
-      "loss": 0.0854,
+      "loss": 0.0194,
       "step": 10564
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.970683736895816e-07,
-      "loss": 0.0317,
+      "loss": 0.0465,
       "step": 10565
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.952129139994434e-07,
-      "loss": 0.0309,
+      "loss": 0.0609,
       "step": 10566
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.9335745430930514e-07,
-      "loss": 0.0857,
+      "loss": 0.0098,
       "step": 10567
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.915019946191669e-07,
-      "loss": 0.1372,
+      "loss": 0.0166,
       "step": 10568
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.8964653492902867e-07,
-      "loss": 0.0319,
+      "loss": 0.1125,
       "step": 10569
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.877910752388905e-07,
-      "loss": 0.0861,
+      "loss": 0.0781,
       "step": 10570
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.8593561554875226e-07,
-      "loss": 0.0844,
+      "loss": 0.0489,
       "step": 10571
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.84080155858614e-07,
-      "loss": 0.1349,
+      "loss": 0.0212,
       "step": 10572
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.822246961684758e-07,
-      "loss": 0.1919,
+      "loss": 0.035,
       "step": 10573
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.8036923647833755e-07,
-      "loss": 0.0845,
+      "loss": 0.051,
       "step": 10574
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.785137767881993e-07,
-      "loss": 0.0828,
+      "loss": 0.045,
       "step": 10575
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.766583170980611e-07,
-      "loss": 0.1374,
+      "loss": 0.1049,
       "step": 10576
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.7480285740792284e-07,
-      "loss": 0.0318,
+      "loss": 0.0226,
       "step": 10577
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.7294739771778466e-07,
-      "loss": 0.0312,
+      "loss": 0.1186,
       "step": 10578
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.710919380276464e-07,
-      "loss": 0.1381,
+      "loss": 0.0648,
       "step": 10579
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.6923647833750814e-07,
-      "loss": 0.2447,
+      "loss": 0.1213,
       "step": 10580
     },
     {
       "epoch": 2.94,
       "learning_rate": 3.673810186473699e-07,
-      "loss": 0.1404,
+      "loss": 0.0656,
       "step": 10581
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.6552555895723167e-07,
-      "loss": 0.2975,
+      "loss": 0.0427,
       "step": 10582
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.6367009926709343e-07,
-      "loss": 0.1425,
+      "loss": 0.0291,
       "step": 10583
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.618146395769552e-07,
-      "loss": 0.0877,
+      "loss": 0.0701,
       "step": 10584
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.5995917988681696e-07,
-      "loss": 0.0884,
+      "loss": 0.022,
       "step": 10585
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.581037201966787e-07,
-      "loss": 0.0853,
+      "loss": 0.2428,
       "step": 10586
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.562482605065405e-07,
-      "loss": 0.1925,
+      "loss": 0.1191,
       "step": 10587
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.543928008164023e-07,
-      "loss": 0.1878,
+      "loss": 0.0103,
       "step": 10588
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.5253734112626407e-07,
-      "loss": 0.1968,
+      "loss": 0.0939,
       "step": 10589
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.5068188143612584e-07,
-      "loss": 0.1905,
+      "loss": 0.0311,
       "step": 10590
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.488264217459876e-07,
-      "loss": 0.085,
+      "loss": 0.0474,
       "step": 10591
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.4697096205584937e-07,
-      "loss": 0.0309,
+      "loss": 0.1411,
       "step": 10592
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.4511550236571113e-07,
-      "loss": 0.1949,
+      "loss": 0.0058,
       "step": 10593
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.432600426755729e-07,
-      "loss": 0.2462,
+      "loss": 0.0081,
       "step": 10594
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.414045829854347e-07,
-      "loss": 0.1935,
+      "loss": 0.0755,
       "step": 10595
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.395491232952965e-07,
-      "loss": 0.1931,
+      "loss": 0.0138,
       "step": 10596
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.3769366360515824e-07,
-      "loss": 0.0312,
+      "loss": 0.1377,
       "step": 10597
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.3583820391502e-07,
-      "loss": 0.2447,
+      "loss": 0.206,
       "step": 10598
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.339827442248817e-07,
-      "loss": 0.1891,
+      "loss": 0.0597,
       "step": 10599
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.321272845347435e-07,
-      "loss": 0.1392,
+      "loss": 0.0772,
       "step": 10600
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.3027182484460525e-07,
-      "loss": 0.1397,
+      "loss": 0.0191,
       "step": 10601
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.28416365154467e-07,
-      "loss": 0.143,
+      "loss": 0.0642,
       "step": 10602
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.265609054643288e-07,
-      "loss": 0.0843,
+      "loss": 0.0184,
       "step": 10603
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.2470544577419054e-07,
-      "loss": 0.2484,
+      "loss": 0.0697,
       "step": 10604
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.2284998608405236e-07,
-      "loss": 0.0849,
+      "loss": 0.0437,
       "step": 10605
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.209945263939141e-07,
-      "loss": 0.084,
+      "loss": 0.0745,
       "step": 10606
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.191390667037759e-07,
-      "loss": 0.0308,
+      "loss": 0.0792,
       "step": 10607
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.1728360701363766e-07,
-      "loss": 0.0854,
+      "loss": 0.1062,
       "step": 10608
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.154281473234994e-07,
-      "loss": 0.0319,
+      "loss": 0.019,
       "step": 10609
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.135726876333612e-07,
-      "loss": 0.1388,
+      "loss": 0.0487,
       "step": 10610
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.1171722794322295e-07,
-      "loss": 0.1904,
+      "loss": 0.0488,
       "step": 10611
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.0986176825308477e-07,
-      "loss": 0.1424,
+      "loss": 0.0626,
       "step": 10612
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.080063085629465e-07,
-      "loss": 0.138,
+      "loss": 0.0424,
       "step": 10613
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.0615084887280824e-07,
-      "loss": 0.0805,
+      "loss": 0.071,
       "step": 10614
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.0429538918267e-07,
-      "loss": 0.0863,
+      "loss": 0.0664,
       "step": 10615
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.0243992949253177e-07,
-      "loss": 0.0832,
+      "loss": 0.0817,
       "step": 10616
     },
     {
       "epoch": 2.95,
       "learning_rate": 3.005844698023936e-07,
-      "loss": 0.1381,
+      "loss": 0.0489,
       "step": 10617
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.9872901011225536e-07,
-      "loss": 0.1913,
+      "loss": 0.0146,
       "step": 10618
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.968735504221171e-07,
-      "loss": 0.0849,
+      "loss": 0.0471,
       "step": 10619
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.950180907319789e-07,
-      "loss": 0.1409,
+      "loss": 0.0713,
       "step": 10620
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.9316263104184065e-07,
-      "loss": 0.1879,
+      "loss": 0.0117,
       "step": 10621
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.913071713517024e-07,
-      "loss": 0.0858,
+      "loss": 0.1375,
       "step": 10622
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.894517116615642e-07,
-      "loss": 0.1392,
+      "loss": 0.0177,
       "step": 10623
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.8759625197142594e-07,
-      "loss": 0.1348,
+      "loss": 0.0859,
       "step": 10624
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.857407922812877e-07,
-      "loss": 0.0836,
+      "loss": 0.0833,
       "step": 10625
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.8388533259114947e-07,
-      "loss": 0.0306,
+      "loss": 0.0538,
       "step": 10626
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.8202987290101124e-07,
-      "loss": 0.0867,
+      "loss": 0.0103,
       "step": 10627
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.80174413210873e-07,
-      "loss": 0.0309,
+      "loss": 0.0237,
       "step": 10628
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.7831895352073477e-07,
-      "loss": 0.1408,
+      "loss": 0.1,
       "step": 10629
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.764634938305966e-07,
-      "loss": 0.1399,
+      "loss": 0.1174,
       "step": 10630
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.7460803414045835e-07,
-      "loss": 0.0873,
+      "loss": 0.1109,
       "step": 10631
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.7275257445032006e-07,
-      "loss": 0.1996,
+      "loss": 0.0426,
       "step": 10632
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.708971147601818e-07,
-      "loss": 0.248,
+      "loss": 0.0626,
       "step": 10633
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.690416550700436e-07,
-      "loss": 0.194,
+      "loss": 0.1537,
       "step": 10634
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.671861953799054e-07,
-      "loss": 0.0845,
+      "loss": 0.0131,
       "step": 10635
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.6533073568976717e-07,
-      "loss": 0.1931,
+      "loss": 0.0104,
       "step": 10636
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.6347527599962894e-07,
-      "loss": 0.0311,
+      "loss": 0.0148,
       "step": 10637
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.616198163094907e-07,
-      "loss": 0.1894,
+      "loss": 0.043,
       "step": 10638
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.5976435661935247e-07,
-      "loss": 0.1423,
+      "loss": 0.0503,
       "step": 10639
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.5790889692921423e-07,
-      "loss": 0.084,
+      "loss": 0.1547,
       "step": 10640
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.56053437239076e-07,
-      "loss": 0.1353,
+      "loss": 0.0574,
       "step": 10641
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.5419797754893776e-07,
-      "loss": 0.0859,
+      "loss": 0.0262,
       "step": 10642
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.523425178587995e-07,
-      "loss": 0.0847,
+      "loss": 0.0869,
       "step": 10643
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.504870581686613e-07,
-      "loss": 0.0827,
+      "loss": 0.1493,
       "step": 10644
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.4863159847852305e-07,
-      "loss": 0.0875,
+      "loss": 0.1046,
       "step": 10645
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.467761387883848e-07,
-      "loss": 0.1384,
+      "loss": 0.0854,
       "step": 10646
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.4492067909824664e-07,
-      "loss": 0.0309,
+      "loss": 0.1379,
       "step": 10647
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.430652194081084e-07,
-      "loss": 0.0843,
+      "loss": 0.0762,
       "step": 10648
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.4120975971797017e-07,
-      "loss": 0.191,
+      "loss": 0.0107,
       "step": 10649
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.3935430002783193e-07,
-      "loss": 0.0864,
+      "loss": 0.1028,
       "step": 10650
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.3749884033769367e-07,
-      "loss": 0.1397,
+      "loss": 0.0118,
       "step": 10651
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.3564338064755543e-07,
-      "loss": 0.0307,
+      "loss": 0.0441,
       "step": 10652
     },
     {
       "epoch": 2.96,
       "learning_rate": 2.337879209574172e-07,
-      "loss": 0.0855,
+      "loss": 0.1073,
       "step": 10653
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.31932461267279e-07,
-      "loss": 0.2473,
+      "loss": 0.2426,
       "step": 10654
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.3007700157714075e-07,
-      "loss": 0.1385,
+      "loss": 0.008,
       "step": 10655
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.2822154188700252e-07,
-      "loss": 0.186,
+      "loss": 0.0795,
       "step": 10656
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.2636608219686428e-07,
-      "loss": 0.0859,
+      "loss": 0.0351,
       "step": 10657
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.2451062250672608e-07,
-      "loss": 0.1945,
+      "loss": 0.1286,
       "step": 10658
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.2265516281658784e-07,
-      "loss": 0.1362,
+      "loss": 0.0853,
       "step": 10659
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.207997031264496e-07,
-      "loss": 0.299,
+      "loss": 0.0499,
       "step": 10660
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.1894424343631134e-07,
-      "loss": 0.0834,
+      "loss": 0.2199,
       "step": 10661
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.1708878374617313e-07,
-      "loss": 0.1409,
+      "loss": 0.1117,
       "step": 10662
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.152333240560349e-07,
-      "loss": 0.1967,
+      "loss": 0.0146,
       "step": 10663
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.1337786436589666e-07,
-      "loss": 0.0304,
+      "loss": 0.0919,
       "step": 10664
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.1152240467575843e-07,
-      "loss": 0.0309,
+      "loss": 0.0846,
       "step": 10665
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.0966694498562022e-07,
-      "loss": 0.1388,
+      "loss": 0.0764,
       "step": 10666
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.0781148529548198e-07,
-      "loss": 0.1405,
+      "loss": 0.0385,
       "step": 10667
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.0595602560534375e-07,
-      "loss": 0.0854,
+      "loss": 0.0624,
       "step": 10668
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.041005659152055e-07,
-      "loss": 0.0854,
+      "loss": 0.2446,
       "step": 10669
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.0224510622506725e-07,
-      "loss": 0.1919,
+      "loss": 0.1337,
       "step": 10670
     },
     {
       "epoch": 2.97,
       "learning_rate": 2.0038964653492904e-07,
-      "loss": 0.0312,
+      "loss": 0.0711,
       "step": 10671
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.985341868447908e-07,
-      "loss": 0.1395,
+      "loss": 0.0214,
       "step": 10672
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.9667872715465257e-07,
-      "loss": 0.1935,
+      "loss": 0.0179,
       "step": 10673
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.9482326746451434e-07,
-      "loss": 0.0856,
+      "loss": 0.1166,
       "step": 10674
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.9296780777437613e-07,
-      "loss": 0.1433,
+      "loss": 0.0635,
       "step": 10675
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.911123480842379e-07,
-      "loss": 0.0826,
+      "loss": 0.1135,
       "step": 10676
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.8925688839409966e-07,
-      "loss": 0.0309,
+      "loss": 0.0811,
       "step": 10677
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.8740142870396142e-07,
-      "loss": 0.1903,
+      "loss": 0.1191,
       "step": 10678
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.855459690138232e-07,
-      "loss": 0.2437,
+      "loss": 0.0594,
       "step": 10679
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.8369050932368495e-07,
-      "loss": 0.086,
+      "loss": 0.0877,
       "step": 10680
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.8183504963354672e-07,
-      "loss": 0.25,
+      "loss": 0.0383,
       "step": 10681
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.7997958994340848e-07,
-      "loss": 0.0844,
+      "loss": 0.0592,
       "step": 10682
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.7812413025327025e-07,
-      "loss": 0.2467,
+      "loss": 0.0683,
       "step": 10683
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.7626867056313204e-07,
-      "loss": 0.1364,
+      "loss": 0.0079,
       "step": 10684
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.744132108729938e-07,
-      "loss": 0.197,
+      "loss": 0.1219,
       "step": 10685
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.7255775118285557e-07,
-      "loss": 0.0311,
+      "loss": 0.0742,
       "step": 10686
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.7070229149271736e-07,
-      "loss": 0.1374,
+      "loss": 0.0169,
       "step": 10687
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.6884683180257912e-07,
-      "loss": 0.0311,
+      "loss": 0.0463,
       "step": 10688
     },
     {
       "epoch": 2.97,
       "learning_rate": 1.6699137211244086e-07,
-      "loss": 0.194,
+      "loss": 0.0593,
       "step": 10689
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.6513591242230262e-07,
-      "loss": 0.0861,
+      "loss": 0.0692,
       "step": 10690
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.632804527321644e-07,
-      "loss": 0.0319,
+      "loss": 0.1678,
       "step": 10691
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.6142499304202618e-07,
-      "loss": 0.1397,
+      "loss": 0.0081,
       "step": 10692
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.5956953335188795e-07,
-      "loss": 0.087,
+      "loss": 0.1084,
       "step": 10693
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.577140736617497e-07,
-      "loss": 0.0853,
+      "loss": 0.0407,
       "step": 10694
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.5585861397161147e-07,
-      "loss": 0.1928,
+      "loss": 0.1241,
       "step": 10695
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.5400315428147324e-07,
-      "loss": 0.1961,
+      "loss": 0.0116,
       "step": 10696
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.52147694591335e-07,
-      "loss": 0.1923,
+      "loss": 0.0498,
       "step": 10697
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.502922349011968e-07,
-      "loss": 0.139,
+      "loss": 0.1836,
       "step": 10698
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.4843677521105856e-07,
-      "loss": 0.1452,
+      "loss": 0.0178,
       "step": 10699
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.4658131552092032e-07,
-      "loss": 0.1401,
+      "loss": 0.0262,
       "step": 10700
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.447258558307821e-07,
-      "loss": 0.1344,
+      "loss": 0.0401,
       "step": 10701
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.4287039614064385e-07,
-      "loss": 0.0855,
+      "loss": 0.0308,
       "step": 10702
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.4101493645050562e-07,
-      "loss": 0.1403,
+      "loss": 0.0278,
       "step": 10703
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.3915947676036738e-07,
-      "loss": 0.4137,
+      "loss": 0.1341,
       "step": 10704
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.3730401707022917e-07,
-      "loss": 0.248,
+      "loss": 0.0748,
       "step": 10705
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.354485573800909e-07,
-      "loss": 0.1398,
+      "loss": 0.0427,
       "step": 10706
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.335930976899527e-07,
-      "loss": 0.0829,
+      "loss": 0.0565,
       "step": 10707
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.3173763799981447e-07,
-      "loss": 0.1945,
+      "loss": 0.0175,
       "step": 10708
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.2988217830967623e-07,
-      "loss": 0.1378,
+      "loss": 0.0452,
       "step": 10709
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.28026718619538e-07,
-      "loss": 0.0817,
+      "loss": 0.0189,
       "step": 10710
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.2617125892939976e-07,
-      "loss": 0.2465,
+      "loss": 0.0701,
       "step": 10711
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.2431579923926153e-07,
-      "loss": 0.1958,
+      "loss": 0.1807,
       "step": 10712
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.2246033954912332e-07,
-      "loss": 0.0849,
+      "loss": 0.037,
       "step": 10713
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.2060487985898508e-07,
-      "loss": 0.2468,
+      "loss": 0.0184,
       "step": 10714
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.1874942016884683e-07,
-      "loss": 0.3004,
+      "loss": 0.2282,
       "step": 10715
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.168939604787086e-07,
-      "loss": 0.0858,
+      "loss": 0.0573,
       "step": 10716
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.1503850078857038e-07,
-      "loss": 0.0822,
+      "loss": 0.0774,
       "step": 10717
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.1318304109843214e-07,
-      "loss": 0.0312,
+      "loss": 0.083,
       "step": 10718
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.1132758140829392e-07,
-      "loss": 0.0314,
+      "loss": 0.0422,
       "step": 10719
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.0947212171815567e-07,
-      "loss": 0.1925,
+      "loss": 0.0664,
       "step": 10720
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.0761666202801745e-07,
-      "loss": 0.0311,
+      "loss": 0.1073,
       "step": 10721
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.0576120233787921e-07,
-      "loss": 0.0318,
+      "loss": 0.0655,
       "step": 10722
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.0390574264774099e-07,
-      "loss": 0.2518,
+      "loss": 0.0492,
       "step": 10723
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.0205028295760276e-07,
-      "loss": 0.0863,
+      "loss": 0.0507,
       "step": 10724
     },
     {
       "epoch": 2.98,
       "learning_rate": 1.0019482326746452e-07,
-      "loss": 0.0863,
+      "loss": 0.1453,
       "step": 10725
     },
     {
       "epoch": 2.99,
       "learning_rate": 9.833936357732629e-08,
-      "loss": 0.14,
+      "loss": 0.1869,
       "step": 10726
     },
     {
       "epoch": 2.99,
       "learning_rate": 9.648390388718806e-08,
-      "loss": 0.3062,
+      "loss": 0.109,
       "step": 10727
     },
     {
       "epoch": 2.99,
       "learning_rate": 9.462844419704983e-08,
-      "loss": 0.084,
+      "loss": 0.0667,
       "step": 10728
     },
     {
       "epoch": 2.99,
       "learning_rate": 9.27729845069116e-08,
-      "loss": 0.0866,
+      "loss": 0.1347,
       "step": 10729
     },
     {
       "epoch": 2.99,
       "learning_rate": 9.091752481677336e-08,
-      "loss": 0.1432,
+      "loss": 0.1424,
       "step": 10730
     },
     {
       "epoch": 2.99,
       "learning_rate": 8.906206512663512e-08,
-      "loss": 0.2995,
+      "loss": 0.0464,
       "step": 10731
     },
     {
       "epoch": 2.99,
       "learning_rate": 8.72066054364969e-08,
-      "loss": 0.0869,
+      "loss": 0.0569,
       "step": 10732
     },
     {
       "epoch": 2.99,
       "learning_rate": 8.535114574635868e-08,
-      "loss": 0.0309,
+      "loss": 0.0772,
       "step": 10733
     },
     {
       "epoch": 2.99,
       "learning_rate": 8.349568605622043e-08,
-      "loss": 0.1985,
+      "loss": 0.0413,
       "step": 10734
     },
     {
       "epoch": 2.99,
       "learning_rate": 8.16402263660822e-08,
-      "loss": 0.0865,
+      "loss": 0.0736,
       "step": 10735
     },
     {
       "epoch": 2.99,
       "learning_rate": 7.978476667594397e-08,
-      "loss": 0.1383,
+      "loss": 0.0731,
       "step": 10736
     },
     {
       "epoch": 2.99,
       "learning_rate": 7.792930698580574e-08,
-      "loss": 0.0849,
+      "loss": 0.0232,
       "step": 10737
     },
     {
       "epoch": 2.99,
       "learning_rate": 7.60738472956675e-08,
-      "loss": 0.2452,
+      "loss": 0.0229,
       "step": 10738
     },
     {
       "epoch": 2.99,
       "learning_rate": 7.421838760552928e-08,
-      "loss": 0.1963,
+      "loss": 0.178,
       "step": 10739
     },
     {
       "epoch": 2.99,
       "learning_rate": 7.236292791539104e-08,
-      "loss": 0.1933,
+      "loss": 0.0136,
       "step": 10740
     },
     {
       "epoch": 2.99,
       "learning_rate": 7.050746822525281e-08,
-      "loss": 0.4133,
+      "loss": 0.0528,
       "step": 10741
     },
     {
       "epoch": 2.99,
       "learning_rate": 6.865200853511459e-08,
-      "loss": 0.1396,
+      "loss": 0.0177,
       "step": 10742
     },
     {
       "epoch": 2.99,
       "learning_rate": 6.679654884497635e-08,
-      "loss": 0.1948,
+      "loss": 0.0433,
       "step": 10743
     },
     {
       "epoch": 2.99,
       "learning_rate": 6.494108915483812e-08,
-      "loss": 0.3507,
+      "loss": 0.0569,
       "step": 10744
     },
     {
       "epoch": 2.99,
       "learning_rate": 6.308562946469988e-08,
-      "loss": 0.0303,
+      "loss": 0.0756,
       "step": 10745
     },
     {
       "epoch": 2.99,
       "learning_rate": 6.123016977456166e-08,
-      "loss": 0.2427,
+      "loss": 0.019,
       "step": 10746
     },
     {
       "epoch": 2.99,
       "learning_rate": 5.937471008442342e-08,
-      "loss": 0.3592,
+      "loss": 0.0164,
       "step": 10747
     },
     {
       "epoch": 2.99,
       "learning_rate": 5.751925039428519e-08,
-      "loss": 0.0843,
+      "loss": 0.0975,
       "step": 10748
     },
     {
       "epoch": 2.99,
       "learning_rate": 5.566379070414696e-08,
-      "loss": 0.0307,
+      "loss": 0.1497,
       "step": 10749
     },
     {
       "epoch": 2.99,
       "learning_rate": 5.3808331014008725e-08,
-      "loss": 0.2501,
+      "loss": 0.0612,
       "step": 10750
     },
     {
       "epoch": 2.99,
       "learning_rate": 5.1952871323870496e-08,
-      "loss": 0.1925,
+      "loss": 0.076,
       "step": 10751
     },
     {
       "epoch": 2.99,
       "learning_rate": 5.009741163373226e-08,
-      "loss": 0.2435,
+      "loss": 0.0708,
       "step": 10752
     },
     {
       "epoch": 2.99,
       "learning_rate": 4.824195194359403e-08,
-      "loss": 0.194,
+      "loss": 0.1878,
       "step": 10753
     },
     {
       "epoch": 2.99,
       "learning_rate": 4.63864922534558e-08,
-      "loss": 0.0871,
+      "loss": 0.0579,
       "step": 10754
     },
     {
       "epoch": 2.99,
       "learning_rate": 4.453103256331756e-08,
-      "loss": 0.1937,
+      "loss": 0.028,
       "step": 10755
     },
     {
       "epoch": 2.99,
       "learning_rate": 4.267557287317934e-08,
-      "loss": 0.1955,
+      "loss": 0.084,
       "step": 10756
     },
     {
       "epoch": 2.99,
       "learning_rate": 4.08201131830411e-08,
-      "loss": 0.0854,
+      "loss": 0.0494,
       "step": 10757
     },
     {
       "epoch": 2.99,
       "learning_rate": 3.896465349290287e-08,
-      "loss": 0.0883,
+      "loss": 0.0373,
       "step": 10758
     },
     {
       "epoch": 2.99,
       "learning_rate": 3.710919380276464e-08,
-      "loss": 0.0305,
+      "loss": 0.0497,
       "step": 10759
     },
     {
       "epoch": 2.99,
       "learning_rate": 3.5253734112626405e-08,
-      "loss": 0.1374,
+      "loss": 0.0704,
       "step": 10760
     },
     {
       "epoch": 2.99,
       "learning_rate": 3.3398274422488176e-08,
-      "loss": 0.142,
+      "loss": 0.0517,
       "step": 10761
     },
     {
       "epoch": 3.0,
       "learning_rate": 3.154281473234994e-08,
-      "loss": 0.031,
+      "loss": 0.063,
       "step": 10762
     },
     {
       "epoch": 3.0,
       "learning_rate": 2.968735504221171e-08,
-      "loss": 0.0854,
+      "loss": 0.1295,
       "step": 10763
     },
     {
       "epoch": 3.0,
       "learning_rate": 2.783189535207348e-08,
-      "loss": 0.2395,
+      "loss": 0.0513,
       "step": 10764
     },
     {
       "epoch": 3.0,
       "learning_rate": 2.5976435661935248e-08,
-      "loss": 0.2981,
+      "loss": 0.0592,
       "step": 10765
     },
     {
       "epoch": 3.0,
       "learning_rate": 2.4120975971797016e-08,
-      "loss": 0.1938,
+      "loss": 0.0338,
       "step": 10766
     },
     {
       "epoch": 3.0,
       "learning_rate": 2.226551628165878e-08,
-      "loss": 0.1966,
+      "loss": 0.0112,
       "step": 10767
     },
     {
       "epoch": 3.0,
       "learning_rate": 2.041005659152055e-08,
-      "loss": 0.0858,
+      "loss": 0.092,
       "step": 10768
     },
     {
       "epoch": 3.0,
       "learning_rate": 1.855459690138232e-08,
-      "loss": 0.1933,
+      "loss": 0.1126,
       "step": 10769
     },
     {
       "epoch": 3.0,
       "learning_rate": 1.6699137211244088e-08,
-      "loss": 0.0876,
+      "loss": 0.0537,
       "step": 10770
     },
     {
       "epoch": 3.0,
       "learning_rate": 1.4843677521105854e-08,
-      "loss": 0.1418,
+      "loss": 0.026,
       "step": 10771
     },
     {
       "epoch": 3.0,
       "learning_rate": 1.2988217830967624e-08,
-      "loss": 0.0852,
+      "loss": 0.123,
       "step": 10772
     },
     {
       "epoch": 3.0,
       "learning_rate": 1.113275814082939e-08,
-      "loss": 0.1367,
+      "loss": 0.0397,
       "step": 10773
     },
     {
       "epoch": 3.0,
       "learning_rate": 9.27729845069116e-09,
-      "loss": 0.1403,
+      "loss": 0.0209,
       "step": 10774
     },
     {
       "epoch": 3.0,
       "learning_rate": 7.421838760552927e-09,
-      "loss": 0.1379,
+      "loss": 0.015,
       "step": 10775
     },
     {
       "epoch": 3.0,
       "learning_rate": 5.566379070414695e-09,
-      "loss": 0.0813,
+      "loss": 0.0184,
       "step": 10776
     },
     {
       "epoch": 3.0,
       "learning_rate": 3.7109193802764636e-09,
-      "loss": 0.1317,
+      "loss": 0.1582,
       "step": 10777
     },
     {
       "epoch": 3.0,
       "learning_rate": 1.8554596901382318e-09,
-      "loss": 0.2475,
+      "loss": 0.0178,
       "step": 10778
     },
     {
       "epoch": 3.0,
       "learning_rate": 0.0,
-      "loss": 0.1077,
+      "loss": 0.0138,
       "step": 10779
     },
     {
       "epoch": 3.0,
       "step": 10779,
-      "total_flos": 5.663507435559936e+16,
-      "train_loss": 0.14667468480848764,
-      "train_runtime": 2515.9928,
-      "train_samples_per_second": 274.165,
-      "train_steps_per_second": 4.284
+      "total_flos": 9.061611896895898e+17,
+      "train_loss": 0.08240926686040774,
+      "train_runtime": 7566.0702,
+      "train_samples_per_second": 91.17,
+      "train_steps_per_second": 1.425
     }
   ],
   "max_steps": 10779,
   "num_train_epochs": 3,
-  "total_flos": 5.663507435559936e+16,
+  "total_flos": 9.061611896895898e+17,
   "trial_name": null,
   "trial_params": null
 }